consul/command/debug/debug.go

746 lines
19 KiB
Go
Raw Permalink Normal View History

New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
package debug
import (
"archive/tar"
"compress/gzip"
"encoding/json"
"errors"
"flag"
"fmt"
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
"golang.org/x/sync/errgroup"
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
"io"
"io/ioutil"
"os"
"path/filepath"
"strings"
"time"
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
"github.com/hashicorp/go-multierror"
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
"github.com/mitchellh/cli"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/consul/command/flags"
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
)
const (
// debugInterval is the interval in which to capture dynamic information
// when running debug
debugInterval = 30 * time.Second
// debugDuration is the total duration that debug runs before being
// shut down
debugDuration = 2 * time.Minute
// debugDurationGrace is a period of time added to the specified
// duration to allow intervals to capture within that time
debugDurationGrace = 2 * time.Second
// debugMinInterval is the minimum a user can configure the interval
// to prevent accidental DOS
debugMinInterval = 5 * time.Second
// debugMinDuration is the minimum a user can configure the duration
// to ensure that all information can be collected in time
debugMinDuration = 10 * time.Second
// debugArchiveExtension is the extension for archive files
debugArchiveExtension = ".tar.gz"
// debugProtocolVersion is the version of the package that is
// generated. If this format changes interface, this version
// can be incremented so clients can selectively support packages
debugProtocolVersion = 1
)
func New(ui cli.Ui, shutdownCh <-chan struct{}) *cmd {
ui = &cli.PrefixedUi{
OutputPrefix: "==> ",
InfoPrefix: " ",
ErrorPrefix: "==> ",
Ui: ui,
}
c := &cmd{UI: ui, shutdownCh: shutdownCh}
c.init()
return c
}
type cmd struct {
UI cli.Ui
flags *flag.FlagSet
http *flags.HTTPFlags
help string
shutdownCh <-chan struct{}
// flags
interval time.Duration
duration time.Duration
output string
archive bool
capture []string
client *api.Client
// validateTiming can be used to skip validation of interval, duration. This
// is primarily useful for testing
validateTiming bool
index *debugIndex
}
// debugIndex is used to manage the summary of all data recorded
// during the debug, to be written to json at the end of the run
// and stored at the root. Each attribute corresponds to a file or files.
type debugIndex struct {
// Version of the debug package
Version int
// Version of the target Consul agent
AgentVersion string
Interval string
Duration string
Targets []string
}
func (c *cmd) init() {
c.flags = flag.NewFlagSet("", flag.ContinueOnError)
defaultFilename := fmt.Sprintf("consul-debug-%d", time.Now().Unix())
c.flags.Var((*flags.AppendSliceValue)(&c.capture), "capture",
fmt.Sprintf("One or more types of information to capture. This can be used "+
"to capture a subset of information, and defaults to capturing "+
"everything available. Possible information for capture: %s. "+
"This can be repeated multiple times.", strings.Join(c.defaultTargets(), ", ")))
c.flags.DurationVar(&c.interval, "interval", debugInterval,
fmt.Sprintf("The interval in which to capture dynamic information such as "+
"telemetry, and profiling. Defaults to %s.", debugInterval))
c.flags.DurationVar(&c.duration, "duration", debugDuration,
fmt.Sprintf("The total time to record information. "+
"Defaults to %s.", debugDuration))
c.flags.BoolVar(&c.archive, "archive", true, "Boolean value for if the files "+
"should be archived and compressed. Setting this to false will skip the "+
"archive step and leave the directory of information on the current path.")
c.flags.StringVar(&c.output, "output", defaultFilename, "The path "+
"to the compressed archive that will be created with the "+
"information after collection.")
c.http = &flags.HTTPFlags{}
flags.Merge(c.flags, c.http.ClientFlags())
c.help = flags.Usage(help, c.flags)
c.validateTiming = true
}
func (c *cmd) Run(args []string) int {
if err := c.flags.Parse(args); err != nil {
c.UI.Error(fmt.Sprintf("Error parsing flags: %s", err))
return 1
}
if len(c.flags.Args()) > 0 {
c.UI.Error("debug: Too many arguments provided, expected 0")
return 1
}
// Connect to the agent
client, err := c.http.APIClient()
if err != nil {
c.UI.Error(fmt.Sprintf("Error connecting to Consul agent: %s", err))
return 1
}
c.client = client
version, err := c.prepare()
if err != nil {
c.UI.Error(fmt.Sprintf("Capture validation failed: %v", err))
return 1
}
archiveName := c.output
// Show the user the final file path if archiving
if c.archive {
archiveName = archiveName + debugArchiveExtension
}
c.UI.Output("Starting debugger and capturing static information...")
// Output metadata about target agent
c.UI.Info(fmt.Sprintf(" Agent Version: '%s'", version))
c.UI.Info(fmt.Sprintf(" Interval: '%s'", c.interval))
c.UI.Info(fmt.Sprintf(" Duration: '%s'", c.duration))
c.UI.Info(fmt.Sprintf(" Output: '%s'", archiveName))
c.UI.Info(fmt.Sprintf(" Capture: '%s'", strings.Join(c.capture, ", ")))
// Record some information for the index at the root of the archive
index := &debugIndex{
Version: debugProtocolVersion,
AgentVersion: version,
Interval: c.interval.String(),
Duration: c.duration.String(),
Targets: c.capture,
}
// Add the extra grace period to ensure
// all intervals will be captured within the time allotted
c.duration = c.duration + debugDurationGrace
// Capture static information from the target agent
err = c.captureStatic()
if err != nil {
c.UI.Warn(fmt.Sprintf("Static capture failed: %v", err))
}
// Capture dynamic information from the target agent, blocking for duration
if c.configuredTarget("metrics") || c.configuredTarget("logs") || c.configuredTarget("pprof") {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
g := new(errgroup.Group)
g.Go(c.captureInterval)
g.Go(c.captureLongRunning)
err = g.Wait()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
if err != nil {
c.UI.Error(fmt.Sprintf("Error encountered during collection: %v", err))
}
}
// Write the index document
idxMarshalled, err := json.MarshalIndent(index, "", "\t")
if err != nil {
c.UI.Error(fmt.Sprintf("Error marshalling index document: %v", err))
return 1
}
err = ioutil.WriteFile(fmt.Sprintf("%s/index.json", c.output), idxMarshalled, 0644)
if err != nil {
c.UI.Error(fmt.Sprintf("Error creating index document: %v", err))
return 1
}
// Archive the data if configured to
if c.archive {
err = c.createArchive()
if err != nil {
c.UI.Warn(fmt.Sprintf("Archive creation failed: %v", err))
return 1
}
}
c.UI.Info(fmt.Sprintf("Saved debug archive: %s", archiveName))
return 0
}
// prepare validates agent settings against targets and prepares the environment for capturing
func (c *cmd) prepare() (version string, err error) {
// Ensure realistic duration and intervals exists
if c.validateTiming {
if c.duration < debugMinDuration {
return "", fmt.Errorf("duration must be longer than %s", debugMinDuration)
}
if c.interval < debugMinInterval {
return "", fmt.Errorf("interval must be longer than %s", debugMinDuration)
}
if c.duration < c.interval {
return "", fmt.Errorf("duration (%s) must be longer than interval (%s)", c.duration, c.interval)
}
}
// Retrieve and process agent information necessary to validate
self, err := c.client.Agent().Self()
if err != nil {
return "", fmt.Errorf("error querying target agent: %s. verify connectivity and agent address", err)
}
version, ok := self["Config"]["Version"].(string)
if !ok {
return "", fmt.Errorf("agent response did not contain version key")
}
// If none are specified we will collect information from
// all by default
if len(c.capture) == 0 {
c.capture = c.defaultTargets()
}
for _, t := range c.capture {
if !c.allowedTarget(t) {
return version, fmt.Errorf("target not found: %s", t)
}
}
if _, err := os.Stat(c.output); os.IsNotExist(err) {
err := os.MkdirAll(c.output, 0755)
if err != nil {
return version, fmt.Errorf("could not create output directory: %s", err)
}
} else {
return version, fmt.Errorf("output directory already exists: %s", c.output)
}
return version, nil
}
// captureStatic captures static target information and writes it
// to the output path
func (c *cmd) captureStatic() error {
// Collect errors via multierror as we want to gracefully
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
// fail if an API is inaccessible
var errs error
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
// Collect the named outputs here
outputs := make(map[string]interface{})
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
// Capture host information
if c.configuredTarget("host") {
host, err := c.client.Agent().Host()
if err != nil {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
errs = multierror.Append(errs, err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
outputs["host"] = host
}
// Capture agent information
if c.configuredTarget("agent") {
agent, err := c.client.Agent().Self()
if err != nil {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
errs = multierror.Append(errs, err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
outputs["agent"] = agent
}
// Capture cluster members information, including WAN
if c.configuredTarget("cluster") {
members, err := c.client.Agent().Members(true)
if err != nil {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
errs = multierror.Append(errs, err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
outputs["cluster"] = members
}
// Write all outputs to disk as JSON
for output, v := range outputs {
marshaled, err := json.MarshalIndent(v, "", "\t")
if err != nil {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
errs = multierror.Append(errs, err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
err = ioutil.WriteFile(fmt.Sprintf("%s/%s.json", c.output, output), marshaled, 0644)
if err != nil {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
errs = multierror.Append(errs, err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
}
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
return errs
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
// captureInterval blocks for the duration of the command
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
// specified by the duration flag, capturing the dynamic
// targets at the interval specified
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
func (c *cmd) captureInterval() error {
intervalChn := time.NewTicker(c.interval)
defer intervalChn.Stop()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
durationChn := time.After(c.duration)
intervalCount := 0
c.UI.Output(fmt.Sprintf("Beginning capture interval %s (%d)", time.Now().Local().String(), intervalCount))
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
err := captureShortLived(c)
if err != nil {
return err
}
c.UI.Output(fmt.Sprintf("Capture successful %s (%d)", time.Now().Local().String(), intervalCount))
for {
select {
case t := <-intervalChn.C:
intervalCount++
err := captureShortLived(c)
if err != nil {
return err
}
c.UI.Output(fmt.Sprintf("Capture successful %s (%d)", t.Local().String(), intervalCount))
case <-durationChn:
intervalChn.Stop()
return nil
case <-c.shutdownCh:
return errors.New("stopping collection due to shutdown signal")
}
}
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
func captureShortLived(c *cmd) error {
g := new(errgroup.Group)
timestamp := time.Now().Local().Unix()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
timestampDir, err := c.createTimestampDir(timestamp)
if err != nil {
return err
}
if c.configuredTarget("pprof") {
g.Go(func() error {
return c.captureHeap(timestampDir)
})
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
g.Go(func() error {
return c.captureGoRoutines(timestampDir)
})
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
// Capture metrics
if c.configuredTarget("metrics") {
g.Go(func() error {
return c.captureMetrics(timestampDir)
})
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
return g.Wait()
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
func (c *cmd) createTimestampDir(timestamp int64) (string, error) {
// Make the directory that will store all captured data
// for this interval
timestampDir := fmt.Sprintf("%s/%d", c.output, timestamp)
err := os.MkdirAll(timestampDir, 0755)
if err != nil {
return "", err
}
return timestampDir, nil
}
func (c *cmd) captureLongRunning() error {
timestamp := time.Now().Local().Unix()
timestampDir, err := c.createTimestampDir(timestamp)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
if err != nil {
return err
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
g := new(errgroup.Group)
// Capture a profile/trace with a minimum of 1s
s := c.duration.Seconds()
if s < 1 {
s = 1
}
// Capture pprof
if c.configuredTarget("pprof") {
g.Go(func() error {
return c.captureProfile(s, timestampDir)
})
g.Go(func() error {
return c.captureTrace(s, timestampDir)
})
}
// Capture logs
if c.configuredTarget("logs") {
g.Go(func() error {
return c.captureLogs(timestampDir)
})
}
return g.Wait()
}
func (c *cmd) captureGoRoutines(timestampDir string) error {
gr, err := c.client.Debug().Goroutine()
if err != nil {
return fmt.Errorf("failed to collect goroutine profile: %w", err)
}
err = ioutil.WriteFile(fmt.Sprintf("%s/goroutine.prof", timestampDir), gr, 0644)
return err
}
func (c *cmd) captureTrace(s float64, timestampDir string) error {
trace, err := c.client.Debug().Trace(int(s))
if err != nil {
return fmt.Errorf("failed to collect trace: %w", err)
}
err = ioutil.WriteFile(fmt.Sprintf("%s/trace.out", timestampDir), trace, 0644)
return err
}
func (c *cmd) captureProfile(s float64, timestampDir string) error {
prof, err := c.client.Debug().Profile(int(s))
if err != nil {
return fmt.Errorf("failed to collect cpu profile: %w", err)
}
err = ioutil.WriteFile(fmt.Sprintf("%s/profile.prof", timestampDir), prof, 0644)
return err
}
func (c *cmd) captureHeap(timestampDir string) error {
heap, err := c.client.Debug().Heap()
if err != nil {
return fmt.Errorf("failed to collect heap profile: %w", err)
}
err = ioutil.WriteFile(fmt.Sprintf("%s/heap.prof", timestampDir), heap, 0644)
return err
}
func (c *cmd) captureLogs(timestampDir string) error {
endLogChn := make(chan struct{})
timeIsUp := time.After(c.duration)
logCh, err := c.client.Agent().Monitor("DEBUG", endLogChn, nil)
if err != nil {
return err
}
// Close the log stream
defer close(endLogChn)
// Create the log file for writing
f, err := os.Create(fmt.Sprintf("%s/%s", timestampDir, "consul.log"))
if err != nil {
return err
}
defer f.Close()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
for {
select {
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
case log := <-logCh:
if log == "" {
return nil
}
if _, err = f.WriteString(log + "\n"); err != nil {
return err
}
case <-timeIsUp:
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
return nil
}
}
}
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
func (c *cmd) captureMetrics(timestampDir string) error {
metrics, err := c.client.Agent().Metrics()
if err != nil {
return err
}
marshaled, err := json.MarshalIndent(metrics, "", "\t")
if err != nil {
return err
}
err = ioutil.WriteFile(fmt.Sprintf("%s/%s.json", timestampDir, "metrics"), marshaled, 0644)
return err
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
// allowedTarget returns a boolean if the target is able to be captured
func (c *cmd) allowedTarget(target string) bool {
for _, dt := range c.defaultTargets() {
if dt == target {
return true
}
}
return false
}
// configuredTarget returns a boolean if the target is configured to be
// captured in the command
func (c *cmd) configuredTarget(target string) bool {
for _, dt := range c.capture {
if dt == target {
return true
}
}
return false
}
// createArchive walks the files in the temporary directory
// and creates a tar file that is gzipped with the contents
func (c *cmd) createArchive() error {
path := c.output + debugArchiveExtension
tempName, err := c.createArchiveTemp(path)
if err != nil {
return err
}
if err := os.Rename(tempName, path); err != nil {
return err
}
// fsync the dir to make the rename stick
if err := syncParentDir(path); err != nil {
return err
}
// Remove directory that has been archived
if err := os.RemoveAll(c.output); err != nil {
return fmt.Errorf("failed to remove archived directory: %s", err)
}
return nil
}
func syncParentDir(name string) error {
f, err := os.Open(filepath.Dir(name))
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
if err != nil {
return err
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
defer f.Close()
return f.Sync()
}
func (c *cmd) createArchiveTemp(path string) (tempName string, err error) {
dir := filepath.Dir(path)
name := filepath.Base(path)
f, err := ioutil.TempFile(dir, name+".tmp")
if err != nil {
return "", fmt.Errorf("failed to create compressed temp archive: %s", err)
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
g := gzip.NewWriter(f)
t := tar.NewWriter(g)
tempName = f.Name()
cleanup := func(err error) (string, error) {
_ = t.Close()
_ = g.Close()
_ = f.Close()
_ = os.Remove(tempName)
return "", err
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
err = filepath.Walk(c.output, func(file string, fi os.FileInfo, err error) error {
if err != nil {
return fmt.Errorf("failed to walk filepath for archive: %s", err)
}
header, err := tar.FileInfoHeader(fi, fi.Name())
if err != nil {
return fmt.Errorf("failed to create compressed archive header: %s", err)
}
header.Name = filepath.Join(filepath.Base(c.output), strings.TrimPrefix(file, c.output))
if err := t.WriteHeader(header); err != nil {
return fmt.Errorf("failed to write compressed archive header: %s", err)
}
// Only copy files
if !fi.Mode().IsRegular() {
return nil
}
f, err := os.Open(file)
if err != nil {
return fmt.Errorf("failed to open target files for archive: %s", err)
}
if _, err := io.Copy(t, f); err != nil {
return fmt.Errorf("failed to copy files for archive: %s", err)
}
generate a single debug file for a long duration capture (#10279) * debug: remove the CLI check for debug_enabled The API allows collecting profiles even debug_enabled=false as long as ACLs are enabled. Remove this check from the CLI so that users do not need to set debug_enabled=true for no reason. Also: - fix the API client to return errors on non-200 status codes for debug endpoints - improve the failure messages when pprof data can not be collected Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com> * remove parallel test runs parallel runs create a race condition that fail the debug tests * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * snapshot the timestamp at the beginning of the capture - timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures - capture append to the file if it already exist * Revert "snapshot the timestamp at the beginning of the capture" This reverts commit c2d03346 * Refactor captureDynamic to extract capture logic for each item in a different func * extract wait group outside the go routine to avoid a race condition * capture pprof in a separate go routine * perform a single capture for pprof data for the whole duration * add missing vendor dependency * add a change log and fix documentation to reflect the change * create function for timestamp dir creation and simplify error handling * use error groups and ticker to simplify interval capture loop * Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval * refactor Logs capture routine and add log capture specific test * improve error reporting when log test fail * change test duration to 1s * make time parsing in log line more robust * refactor log time format in a const * test on log line empty the earliest possible and return Co-authored-by: Freddy <freddygv@users.noreply.github.com> * rename function to captureShortLived * more specific changelog Co-authored-by: Paul Banks <banks@banksco.de> * update documentation to reflect current implementation * add test for behavior when invalid param is passed to the command * fix argument line in test * a more detailed description of the new behaviour Co-authored-by: Paul Banks <banks@banksco.de> * print success right after the capture is done * remove an unnecessary error check Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> * upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22 Co-authored-by: Daniel Nephin <dnephin@hashicorp.com> Co-authored-by: Freddy <freddygv@users.noreply.github.com> Co-authored-by: Paul Banks <banks@banksco.de>
2021-06-07 17:00:51 +00:00
return f.Close()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
})
if err != nil {
return cleanup(fmt.Errorf("failed to walk output path for archive: %s", err))
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
// Explicitly close things in the correct order (tar then gzip) so we
// know if they worked.
if err := t.Close(); err != nil {
return cleanup(err)
}
if err := g.Close(); err != nil {
return cleanup(err)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
// Guarantee that the contents of the temp file are flushed to disk.
if err := f.Sync(); err != nil {
return cleanup(err)
}
// Close the temp file and go back to the wrapper function for the rest.
if err := f.Close(); err != nil {
return cleanup(err)
}
return tempName, nil
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
// defaultTargets specifies the list of all targets that
// will be captured by default
func (c *cmd) defaultTargets() []string {
return append(c.dynamicTargets(), c.staticTargets()...)
}
// dynamicTargets returns all the supported targets
// that are retrieved at the interval specified
func (c *cmd) dynamicTargets() []string {
return []string{"metrics", "logs", "pprof"}
}
// staticTargets returns all the supported targets
// that are retrieved at the start of the command execution
func (c *cmd) staticTargets() []string {
return []string{"host", "agent", "cluster"}
}
func (c *cmd) Synopsis() string {
return synopsis
}
func (c *cmd) Help() string {
return c.help
}
const synopsis = "Records a debugging archive for operators"
const help = `
Usage: consul debug [options]
Monitors a Consul agent for the specified period of time, recording
information about the agent, cluster, and environment to an archive
written to the specified path.
If ACLs are enabled, an 'operator:read' token must be supplied in order
to perform this operation.
To create a debug archive in the current directory for the default
duration and interval, capturing all information available:
$ consul debug
The command stores captured data at the configured output path
through the duration, and will archive the data at the same
path if interrupted.
Flags can be used to customize the duration and interval of the
operation. Duration is the total time to capture data for from the target
agent and interval controls how often dynamic data such as metrics
are scraped.
$ consul debug -interval=20s -duration=1m
The capture flag can be specified multiple times to limit information
retrieved.
$ consul debug -capture metrics -capture agent
By default, the archive containing the debugging information is
saved to the current directory as a .tar.gz file. The
output path can be specified, as well as an option to disable
archiving, leaving the directory intact.
$ consul debug -output=/foo/bar/my-debugging -archive=false
Note: Information collected by this command has the potential
to be highly sensitive. Sensitive material such as ACL tokens and
other commonly secret material are redacted automatically, but we
strongly recommend review of the data within the archive prior to
transmitting it.
For a full list of options and examples, please see the Consul
documentation.
`