* Add a test to reproduce the race condition
* Fix race condition by publishing the event after the commit and adding a lock to prevent out of order events.
* split publish to generate the list of events before committing the transaction.
* add changelog
* remove extra func
* Apply suggestions from code review
Co-authored-by: Dan Upton <daniel@floppy.co>
* add comment to explain test
---------
Co-authored-by: Dan Upton <daniel@floppy.co>
Prior to this change, peer services would be targeted by service-default
overrides as long as the new `peer` field was not found in the config entry.
This commit removes that deprecated backwards-compatibility behavior. Now
it is necessary to specify the `peer` field in order for upstream overrides
to apply to a peer upstream.
Currently, if an acceptor peer deletes a peering the dialer's peering
will eventually get to a "terminated" state. If the two clusters need to
be re-peered the acceptor will re-generate the token but the dialer will
encounter this error on the call to establish:
"failed to get addresses to dial peer: failed to refresh peer server
addresses, will continue to use initial addresses: there is no active
peering for "<<<ID>>>""
This is because in `exchangeSecret().GetDialAddresses()` we will get an
error if fetching addresses for an inactive peering. The peering shows
up as inactive at this point because of the existing terminated state.
Rather than checking whether a peering is active we can instead check
whether it was deleted. This way users do not need to delete terminated
peerings in the dialing cluster before re-establishing them.
* Use merge of enterprise meta's rather than new custom method
* Add merge logic for tcp routes
* Add changelog
* Normalize certificate refs on gateways
* Fix infinite call loop
* Explicitly call enterprise meta
This commit swaps the partition field to the local partition for
discovery chains targeting peers. Prior to this change, peer upstreams
would always use a value of default regardless of which partition they
exist in. This caused several issues in xds / proxycfg because of id
mismatches.
Some prior fixes were made to deal with one-off id mismatches that this
PR also cleans up, since they are no longer needed.
* add snapshot restore test
* add logstore as test parameter
* Use the correct image version
* make sure we read the logs from a followers to test the follower snapshot install path.
* update to raf-wal v0.3.0
* add changelog.
* updating changelog for bug description and removed integration test.
* setting up test container builder to only set logStore for 1.15 and higher
---------
Co-authored-by: Paul Banks <pbanks@hashicorp.com>
Co-authored-by: John Murret <john.murret@hashicorp.com>
This commit fixes an issue where trust bundles could not be read
by services in a non-default namespace, unless they had excessive
ACL permissions given to them.
Prior to this change, `service:write` was required in the default
namespace in order to read the trust bundle. Now, `service:write`
to a service in any namespace is sufficient.
If a CA config update did not cause a root change, the codepath would return early and skip some steps which preserve its intermediate certificates and signing key ID. This commit re-orders some code and prevents updates from generating new intermediate certificates.
Co-authored-by: Ashvitha Sridharan <ashvitha.sridharan@hashicorp.com>
Co-authored-by: Freddy <freddygv@users.noreply.github.com>
Add a new envoy flag: "envoy_hcp_metrics_bind_socket_dir", a directory
where a unix socket will be created with the name
`<namespace>_<proxy_id>.sock` to forward Envoy metrics.
If set, this will configure:
- In bootstrap configuration a local stats_sink and static cluster.
These will forward metrics to a loopback listener sent over xDS.
- A dynamic listener listening at the socket path that the previously
defined static cluster is sending metrics to.
- A dynamic cluster that will forward traffic received at this listener
to the hcp-metrics-collector service.
Reasons for having a static cluster pointing at a dynamic listener:
- We want to secure the metrics stream using TLS, but the stats sink can
only be defined in bootstrap config. With dynamic listeners/clusters
we can use the proxy's leaf certificate issued by the Connect CA,
which isn't available at bootstrap time.
- We want to intelligently route to the HCP collector. Configuring its
addreess at bootstrap time limits our flexibility routing-wise. More
on this below.
Reasons for defining the collector as an upstream in `proxycfg`:
- The HCP collector will be deployed as a mesh service.
- Certificate management is taken care of, as mentioned above.
- Service discovery and routing logic is automatically taken care of,
meaning that no code changes are required in the xds package.
- Custom routing rules can be added for the collector using discovery
chain config entries. Initially the collector is expected to be
deployed to each admin partition, but in the future could be deployed
centrally in the default partition. These config entries could even be
managed by HCP itself.
Add support for using existing vault auto-auth configurations as the
provider configuration when using Vault's CA provider with AliCloud.
AliCloud requires 2 extra fields to enable it to use STS (it's preferred
auth setup). Our vault-plugin-auth-alicloud package contained a method
to help generate them as they require you to make an http call to
a faked endpoint proxy to get them (url and headers base64 encoded).
* Add some basic ui improvements for api-gateway services
* Add changelog entry
* Use ternary for null check
* Update gateway doc links
* rename changelog entry for new PR
* Fix test
Receiving an "acl not found" error from an RPC in the agent cache and the
streaming/event components will cause any request loops to cease under the
assumption that they will never work again if the token was destroyed. This
prevents log spam (#14144, #9738).
Unfortunately due to things like:
- authz requests going to stale servers that may not have witnessed the token
creation yet
- authz requests in a secondary datacenter happening before the tokens get
replicated to that datacenter
- authz requests from a primary TO a secondary datacenter happening before the
tokens get replicated to that datacenter
The caller will get an "acl not found" *before* the token exists, rather than
just after. The machinery added above in the linked PRs will kick in and
prevent the request loop from looping around again once the tokens actually
exist.
For `consul-dataplane` usages, where xDS is served by the Consul servers
rather than the clients ultimately this is not a problem because in that
scenario the `agent/proxycfg` machinery is on-demand and launched by a new xDS
stream needing data for a specific service in the catalog. If the watching
goroutines are terminated it ripples down and terminates the xDS stream, which
CDP will eventually re-establish and restart everything.
For Consul client usages, the `agent/proxycfg` machinery is ahead-of-time
launched at service registration time (called "local" in some of the proxycfg
machinery) so when the xDS stream comes in the data is already ready to go. If
the watching goroutines terminate it should terminate the xDS stream, but
there's no mechanism to re-spawn the watching goroutines. If the xDS stream
reconnects it will see no `ConfigSnapshot` and will not get one again until
the client agent is restarted, or the service is re-registered with something
changed in it.
This PR fixes a few things in the machinery:
- there was an inadvertent deadlock in fetching snapshot from the proxycfg
machinery by xDS, such that when the watching goroutine terminated the
snapshots would never be fetched. This caused some of the xDS machinery to
get indefinitely paused and not finish the teardown properly.
- Every 30s we now attempt to re-insert all locally registered services into
the proxycfg machinery.
- When services are re-inserted into the proxycfg machinery we special case
"dead" ones such that we unilaterally replace them rather that doing that
conditionally.
Adds support for the approle auth-method. Only handles using the approle
role/secret to auth and it doesn't support the agent's extra management
configuration options (wrap and delete after read) as they are not
required as part of the auth (ie. they are vault agent things).
* Fix issue where terminating gateway service resolvers weren't properly cleaned up
* Add integration test for cleaning up resolvers
* Add changelog entry
* Use state test and drop integration test
* Leverage ServiceResolver ConnectTimeout for route timeouts to make TerminatingGateway upstream timeouts configurable
* Regenerate golden files
* Add RequestTimeout field
* Add changelog entry
Adds support for a jwt token in a file. Simply reads the file and sends
the read in jwt along to the vault login.
It also supports a legacy mode with the jwt string being passed
directly. In which case the path is made optional.
Does the required dance with the local HTTP endpoint to get the required
data for the jwt based auth setup in Azure. Keeps support for 'legacy'
mode where all login data is passed on via the auth methods parameters.
Refactored check for hardcoded /login fields.
Fixes a regression in #16044
The consul acl token read -self cli command should not require an -accessor-id because typically the persona invoking this would not already know the accessor id of their own token.
Prior to this commit, all peer services were transmitted as connect-enabled
as long as a one or more mesh-gateways were healthy. With this change, there
is now a difference between typical services and connect services transmitted
via peering.
A service will be reported as "connect-enabled" as long as any of these
conditions are met:
1. a connect-proxy sidecar is registered for the service name.
2. a connect-native instance of the service is registered.
3. a service resolver / splitter / router is registered for the service name.
4. a terminating gateway has registered the service.
Fix mesh gateways incorrectly matching peer locality.
This fixes an issue where local mesh gateways use an
incorrect address when attempting to forward traffic to a
peered datacenter. Prior to this change it would use the
lan address instead of the wan if the locality matched. This
should never be done for peering, since we must route all
traffic through the remote mesh gateway.
Prior to this commit, secondary datacenters could not be initialized
as peering acceptors if ACLs were enabled. This is due to the fact that
internal server-to-server API calls would fail because the management
token was not generated. This PR makes it so that both primary and
secondary datacenters generate their own management token whenever
a leader is elected in their respective clusters.
This endpoint shows total services, connect service instances and
billable service instances in the local datacenter or globally. Billable
instances = total service instances - connect services - consul server instances.
* remove legacy tokens
* remove lingering legacy token references from docs
* update language and naming for token secrets and accessor IDs
* updates all tokenID references to clarify accessorID
* remove token type references and lookup tokens by accessorID index
* remove unnecessary constants
* replace additional tokenID param names
* Add warning info for deprecated -id parameter
Co-authored-by: Paul Glass <pglass@hashicorp.com>
* Update field comment
Co-authored-by: Paul Glass <pglass@hashicorp.com>
---------
Co-authored-by: Paul Glass <pglass@hashicorp.com>
* Add Peer field to service-defaults upstream overrides.
* add api changes, compat mode for service default overrides
* Fixes based on testing
---------
Co-authored-by: DanStough <dan.stough@hashicorp.com>
Use only the agent token for deregistration during anti-entropy
The previous behavior had the agent attempt to use the "service" token
(i.e. from the `token` field in a service definition file), and if that
was not set then it would use the agent token.
The previous behavior was problematic because, if the service token had
been deleted, the deregistration request would fail. The agent would
retry the deregistration during each anti-entropy sync, and the
situation would never resolve.
The new behavior is to only/always use the agent token for service and
check deregistration during anti-entropy. This approach is:
* Simpler: No fallback logic to try different tokens
* Faster (slightly): No time spent attempting the service token
* Correct: The agent token is able to deregister services on that
agent's node, because:
* node:write permissions allow deregistration of services/checks on
that node.
* The agent token must have node:write permission, or else the agent
is not be able to (de)register itself into the catalog
Co-authored-by: Vesa Hagström <weeezes@gmail.com>
* remove legacy tokens
* Update test comment
Co-authored-by: Paul Glass <pglass@hashicorp.com>
* fix imports
* update docs for additional CLI changes
* add test case for anonymous token
* set deprecated api fields to json ignore and fix patch errors
* update changelog to breaking-change
* fix import
* update api docs to remove legacy reference
* fix docs nav data
---------
Co-authored-by: Paul Glass <pglass@hashicorp.com>
* Add support for envoy readiness flags
- add flags 'envoy-ready-bind-port` and `envoy-ready-bind-addr` on consul connect envoy to create a ready listener on that address.
Fix configuration merging for implicit tproxy upstreams.
Change the merging logic so that the wildcard upstream has correct proxy-defaults
and service-defaults values combined into it. It did not previously merge all fields,
and the wildcard upstream did not exist unless service-defaults existed (it ignored
proxy-defaults, essentially).
Change the way we fetch upstream configuration in the xDS layer so that it falls back
to the wildcard when no matching upstream is found. This is what allows implicit peer
upstreams to have the correct "merged" config.
Change proxycfg to always watch local mesh gateway endpoints whenever a peer upstream
is found. This simplifies the logic so that we do not have to inspect the "merged"
configuration on peer upstreams to extract the mesh gateway mode.
Previously, we'd begin a session with the xDS concurrency limiter
regardless of whether the proxy was registered in the catalog or in
the server's local agent state.
This caused problems for users who run `consul connect envoy` directly
against a server rather than a client agent, as the server's locally
registered proxies wouldn't be included in the limiter's capacity.
Now, the `ConfigSource` is responsible for beginning the session and we
only do so for services in the catalog.
Fixes: https://github.com/hashicorp/consul/issues/15753
Enforce lowercase peer names.
Prior to this change peer names could be mixed case.
This can cause issues, as peer names are used as DNS labels
in various locations. It also caused issues with envoy
configuration.
Co-authored-by: Jared Kirschner <85913323+jkirschner-hashicorp@users.noreply.github.com>
Co-authored-by: Jeff Boruszak <104028618+boruszak@users.noreply.github.com>
Fix issue where TLS configuration was ignored for unix sockets in consul connect envoy.
Disable xds check on bootstrap mode and change check to warn only.
This is the OSS portion of enterprise PR 3822.
Adds a custom gRPC balancer that replicates the router's server cycling
behavior. Also enables automatic retries for RESOURCE_EXHAUSTED errors,
which we now get for free.
Previously, these endpoints required `service:write` permission on _any_
service as a sort of proxy for "is the caller allowed to participate in
the mesh?".
Now, they're called as part of the process of establishing a server
connection by any consumer of the consul-server-connection-manager
library, which will include non-mesh workloads (e.g. Consul KV as a
storage backend for Vault) as well as ancillary components such as
consul-k8s' acl-init process, which likely won't have `service:write`
permission.
So this commit relaxes those requirements to accept *any* valid ACL token
on the following gRPC endpoints:
- `hashicorp.consul.dataplane.DataplaneService/GetSupportedDataplaneFeatures`
- `hashicorp.consul.serverdiscovery.ServerDiscoveryService/WatchServers`
- `hashicorp.consul.connectca.ConnectCAService/WatchRoots`
Fix agent cache incorrectly notifying unchanged protobufs.
This change fixes a situation where the protobuf private fields
would be read by reflect.DeepEqual() and indicate data was modified.
This resulted in change notifications being fired every time, which
could cause performance problems in proxycfg.
* add functions for returning the max and min Envoy major versions
- added an UnsupportedEnvoyVersions list
- removed an unused error from TestDetermineSupportedProxyFeaturesFromString
- modified minSupportedVersion to use the function for getting the Min Envoy major version. Using just the major version without the patch is equivalent to using `.0`
* added a function for executing the envoy --version command
- added a new exec.go file to not be locked to unix system
* added envoy version check when using consul connect envoy
* added changelog entry
* added docs change
The new balancer is a patched version of gRPC's default pick_first balancer
which removes the behavior of preserving the active subconnection if
a list of new addresses contains the currently active address.
* feat(ingress-gateway): support outlier detection of upstream service for ingress gateway
* changelog
Co-authored-by: Eric Haberkorn <erichaberkorn@gmail.com>
Fix local mesh gateway with peering discovery chains.
Prior to this patch, discovery chains with peers would not
properly honor the mesh gateway mode for two reasons.
1. An incorrect target upstream ID was used to lookup the
mesh gateway mode. To fix this, the parent upstream uid is
now used instead of the discovery-chain-target-uid to find
the intended mesh gateway mode.
2. The watch for local mesh gateways was never initialized
for discovery chains. To fix this, the discovery chains are
now scanned, and a local GW watch is spawned if: the mesh
gateway mode is local and the target is a peering connection.
All of the current integration tests where Vault is the Connect CA now use non-root tokens for the test. This helps us detect privilege changes in the vault model so we can keep our guides up to date.
One larger change was that the RenewIntermediate function got refactored slightly so it could be used from a test, rather than the large duplicated function we were testing in a test which seemed error prone.
The fix outlined and merged in #15253 fixed the issue as it occurs in the primary
DC. There is a similar issue that arises when vault is used as the Connect CA in a
secondary datacenter that is fixed by this PR.
Additionally: this PR adds support to run the existing suite of vault related integration
tests against the last 4 versions of vault (1.9, 1.10, 1.11, 1.12)
* Fixing CLI ACL token processing unexpected precedence
* Minor flow format and add Changelog
* Fixed failed tests and improve error logging message
* Add unit test cases and minor changes from code review
* Unset env var once the test case finishes running
* remove label FINISH
During peer stream replication we flatten checks from the source cluster and build one thin overall check to hide the irrelevant details from the consuming cluster. This flattening logic did correctly flip to non-passing if there were any non-passing checks, but WHICH status it got during that was random (warn/error).
Also it didn't represent "maintenance" operations. There is an api package call AggregatedStatus which more correctly flattened check statuses.
This PR replicated the more complete logic into the peer stream package.
* Remove log line about server mgmt token init
Currently the server management token is only being bootstrapped in the
primary datacenter. That means that servers on the secondary datacenter
will never have this token available, and would log this line any time a
token is resolved.
Bootstrapping the token in secondary datacenters will be done in a
follow-up.
* Add changelog entry
This change was necessary, because the configuration was always
generated with a gRPC TLS port, which did not exist in Consul 1.13,
and would result in the server failing to launch with an error.
This code checks the version of Consul and conditionally adds the
gRPC TLS port, only if the version number is greater than 1.14.
Consul used to rely on implicit issuer selection when calling Vault endpoints to issue new CSRs. Vault 1.11+ changed that behavior, which caused Consul to check the wrong (previous) issuer when renewing its Intermediate CA. This patch allows Consul to explicitly set a default issuer when it detects that the response from Vault is 1.11+.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
* auto-config: relax node name validation for JWT authorization
This changes the JWT authorization logic to allow all non-whitespace,
non-quote characters when validating node names. Consul had previously
allowed these characters in node names, until this validation was added
to fix a security vulnerability with whitespace/quotes being passed to
the `bexpr` library. This unintentionally broke node names with
characters like `.` which aren't related to this vulnerability.
* Update website/content/docs/agent/config/cli-flags.mdx
Co-authored-by: trujillo-adam <47586768+trujillo-adam@users.noreply.github.com>
Co-authored-by: trujillo-adam <47586768+trujillo-adam@users.noreply.github.com>
* add leadership transfer command
* add RPC call test (flaky)
* add missing import
* add changelog
* add command registration
* Apply suggestions from code review
Co-authored-by: Matt Keeler <mkeeler@users.noreply.github.com>
* add the possibility of providing an id to raft leadership transfer. Add few tests.
* delete old file from cherry pick
* rename changelog filename to PR #
* rename changelog and fix import
* fix failing test
* check for OperatorWrite
Co-authored-by: Matt Keeler <mkeeler@users.noreply.github.com>
* rename from leader-transfer to transfer-leader
* remove version check and add test for operator read
* move struct to operator.go
* first pass
* add code for leader transfer in the grpc backend and tests
* wire the http endpoint to the new grpc endpoint
* remove the RPC endpoint
* remove non needed struct
* fix naming
* add mog glue to API
* fix comment
* remove dead code
* fix linter error
* change package name for proto file
* remove error wrapping
* fix failing test
* add command registration
* add grpc service mock tests
* fix receiver to be pointer
* use defined values
Co-authored-by: Matt Keeler <mkeeler@users.noreply.github.com>
* reuse MockAclAuthorizer
* add documentation
* remove usage of external.TokenFromContext
* fix failing tests
* fix proto generation
* Apply suggestions from code review
Co-authored-by: Jared Kirschner <85913323+jkirschner-hashicorp@users.noreply.github.com>
* Apply suggestions from code review
Co-authored-by: Jared Kirschner <85913323+jkirschner-hashicorp@users.noreply.github.com>
* Apply suggestions from code review
Co-authored-by: Jared Kirschner <85913323+jkirschner-hashicorp@users.noreply.github.com>
* Apply suggestions from code review
* add more context in doc for the reason
* Apply suggestions from docs code review
Co-authored-by: Jeff Boruszak <104028618+boruszak@users.noreply.github.com>
* regenerate proto
* fix linter errors
Co-authored-by: github-team-consul-core <github-team-consul-core@hashicorp.com>
Co-authored-by: Matt Keeler <mkeeler@users.noreply.github.com>
Co-authored-by: Jared Kirschner <85913323+jkirschner-hashicorp@users.noreply.github.com>
Co-authored-by: Jeff Boruszak <104028618+boruszak@users.noreply.github.com>
* connect: strip port from DNS SANs for ingress gateway leaf cert
* connect: format DNS SANs in CreateCSR
* connect: Test wildcard case when formatting SANs
Prevent serving TLS via ports.grpc
We remove the ability to run the ports.grpc in TLS mode to avoid
confusion and to simplify configuration. This breaking change
ensures that any user currently using ports.grpc in an encrypted
mode will receive an error message indicating that ports.grpc_tls
must be explicitly used.
The suggested action for these users is to simply swap their ports.grpc
to ports.grpc_tls in the configuration file. If both ports are defined,
or if the user has not configured TLS for grpc, then the error message
will not be printed.
* update go version to 1.18 for api and sdk, go mod tidy
* removes ioutil usage everywhere which was deprecated in go1.16 in favour of io and os packages. Also introduces a lint rule which forbids use of ioutil going forward.
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
* Fix mesh gateway proxy-defaults not affecting upstreams.
* Clarify distinction with upstream settings
Top-level mesh gateway mode in proxy-defaults and service-defaults gets
merged into NodeService.Proxy.MeshGateway, and only gets merged with
the mode attached to an an upstream in proxycfg/xds.
* Fix mgw mode usage for peered upstreams
There were a couple issues with how mgw mode was being handled for
peered upstreams.
For starters, mesh gateway mode from proxy-defaults
and the top-level of service-defaults gets stored in
NodeService.Proxy.MeshGateway, but the upstream watch for peered data
was only considering the mesh gateway config attached in
NodeService.Proxy.Upstreams[i]. This means that applying a mesh gateway
mode via global proxy-defaults or service-defaults on the downstream
would not have an effect.
Separately, transparent proxy watches for peered upstreams didn't
consider mesh gateway mode at all.
This commit addresses the first issue by ensuring that we overlay the
upstream config for peered upstreams as we do for non-peered. The second
issue is addressed by re-using setupWatchesForPeeredUpstream when
handling transparent proxy updates.
Note that for transparent proxies we do not yet support mesh gateway
mode per upstream, so the NodeService.Proxy.MeshGateway mode is used.
* Fix upstream mesh gateway mode handling in xds
This commit ensures that when determining the mesh gateway mode for
peered upstreams we consider the NodeService.Proxy.MeshGateway config as
a baseline.
In absense of this change, setting a mesh gateway mode via
proxy-defaults or the top-level of service-defaults will not have an
effect for peered upstreams.
* Merge service/proxy defaults in cfg resolver
Previously the mesh gateway mode for connect proxies would be
merged at three points:
1. On servers, in ComputeResolvedServiceConfig.
2. On clients, in MergeServiceConfig.
3. On clients, in proxycfg/xds.
The first merge returns a ServiceConfigResponse where there is a
top-level MeshGateway config from proxy/service-defaults, along with
per-upstream config.
The second merge combines per-upstream config specified at the service
instance with per-upstream config specified centrally.
The third merge combines the NodeService.Proxy.MeshGateway
config containing proxy/service-defaults data with the per-upstream
mode. This third merge is easy to miss, which led to peered upstreams
not considering the mesh gateway mode from proxy-defaults.
This commit removes the third merge, and ensures that all mesh gateway
config is available at the upstream. This way proxycfg/xds do not need
to do additional overlays.
* Ensure that proxy-defaults is considered in wc
Upstream defaults become a synthetic Upstream definition under a
wildcard key "*". Now that proxycfg/xds expect Upstream definitions to
have the final MeshGateway values, this commit ensures that values from
proxy-defaults/service-defaults are the default for this synthetic
upstream.
* Add changelog.
Co-authored-by: freddygv <freddy@hashicorp.com>
Re-add ServerExternalAddresses parameter in GenerateToken endpoint
This reverts commit 5e156772f6
and adds extra functionality to support newer peering behaviors.
* ingress-gateways: don't log error when registering gateway
Previously, when an ingress gateway was registered without a
corresponding ingress gateway config entry, an error was logged
because the watch on the config entry returned a nil result.
This is expected so don't log an error.
* config entry: hardcode proxy-defaults name as global
proxy-defaults can only have the name global. Because of this,
we support not even setting the name in the config file:
```
kind = "proxy-defaults"
```
Previously, writing this would result in the output:
```
Config entry written: proxy-defaults/
```
Now it will output:
```
Config entry written: proxy-defaults/global
```
This change follows what was done for the new Mesh config entry.
* autoencrypt: helpful error for clients with wrong dc
If clients have set a different datacenter than the servers they're
connecting with for autoencrypt, give a helpful error message.
This continues the work done in #14908 where a crude solution to prevent a
goroutine leak was implemented. The former code would launch a perpetual
goroutine family every iteration (+1 +1) and the fixed code simply caused a
new goroutine family to first cancel the prior one to prevent the
leak (-1 +1 == 0).
This PR refactors this code completely to:
- make it more understandable
- remove the recursion-via-goroutine strangeness
- prevent unnecessary RPC fetches when the prior one has errored.
The core issue arose from a conflation of the entry.Fetching field to mean:
- there is an RPC (blocking query) in flight right now
- there is a goroutine running to manage the RPC fetch retry loop
The problem is that the goroutine-leak-avoidance check would treat
Fetching like (2), but within the body of a goroutine it would flip that
boolean back to false before the retry sleep. This would cause a new
chain of goroutines to launch which #14908 would correct crudely.
The refactored code uses a plain for-loop and changes the semantics
to track state for "is there a goroutine associated with this cache entry"
instead of the former.
We use a uint64 unique identity per goroutine instead of a boolean so
that any orphaned goroutines can tell when they've been replaced when
the expiry loop deletes a cache entry while the goroutine is still running
and is later replaced.
Fix an issue where rpc_hold_timeout was being used as the timeout for non-blocking queries. Users should be able to tune read timeouts without fiddling with rpc_hold_timeout. A new configuration `rpc_read_timeout` is created.
Refactor some implementation from the original PR 11500 to remove the misleading linkage between RPCInfo's timeout (used to retry in case of certain modes of failures) and the client RPC timeouts.
There is a bug in the error handling code for the Agent cache subsystem discovered:
1. NotifyCallback calls notifyBlockingQuery which calls getWithIndex in
a loop (which backs off on-error up to 1 minute)
2. getWithIndex calls fetch if there’s no valid entry in the cache
3. fetch starts a goroutine which calls Fetch on the cache-type, waits
for a while (again with backoff up to 1 minute for errors) and then
calls fetch to trigger a refresh
The end result being that every 1 minute notifyBlockingQuery spawns an
ancestry of goroutines that essentially lives forever.
This PR ensures that the goroutine started by `fetch` cancels any prior
goroutine spawned by the same line for the same key.
In isolated testing where a cache type was tweaked to indefinitely
error, this patch prevented goroutine counts from skyrocketing.
In practice this was masked by #14956 and was only uncovered fixing the
other bug.
go test ./agent -run TestAgentConnectCALeafCert_goodNotLocal
would fail when only #14956 was fixed.
Adds a user-configurable rate limiter to proxycfg snapshot delivery,
with a default limit of 250 updates per second.
This addresses a problem observed in our load testing of Consul
Dataplane where updating a "global" resource such as a wildcard
intention or the proxy-defaults config entry could starve the Raft or
Memberlist goroutines of CPU time, causing general cluster instability.
Replaces the reflection-based implementation of proxycfg's
ConfigSnapshot.Clone with code generated by deep-copy.
While load testing server-based xDS (for consul-dataplane) we discovered
this method is extremely expensive. The ConfigSnapshot struct, directly
or indirectly, contains a copy of many of the structs in the agent/structs
package, which creates a large graph for copystructure.Copy to traverse
at runtime, on every proxy reconfiguration.
memdb's `WatchCh` method creates a goroutine that will publish to the
returned channel when the watchset is triggered or the given context
is canceled. Although this is called out in its godoc comment, it's
not obvious that this method creates a goroutine who's lifecycle you
need to manage.
In the xDS capacity controller, we were calling `WatchCh` on each
iteration of the control loop, meaning the number of goroutines would
grow on each autopilot event until there was catalog churn.
In the catalog config source, we were calling `WatchCh` with the
background context, meaning that the goroutine would keep running after
the sync loop had terminated.
Adds another datasource for proxycfg.HTTPChecks, for use on server agents. Typically these checks are performed by local client agents and there is no equivalent of this in agentless (where servers configure consul-dataplane proxies).
Hence, the data source is mostly a no-op on servers but in the case where the service is present within the local state, it delegates to the cache data source.
* Move stats.go from grpc-internal to grpc-middleware
* Update grpc server metrics with server type label
* Add stats test to grpc-external
* Remove global metrics instance from grpc server tests
* Configure Envoy alpn_protocols based on service protocol
* define alpnProtocols in a more standard way
* http2 protocol should be h2 only
* formatting
* add test for getAlpnProtocol()
* create changelog entry
* change scope is connect-proxy
* ignore errors on ParseProxyConfig; fixes linter
* add tests for grpc and http2 public listeners
* remove newlines from PR
* Add alpn_protocol configuration for ingress gateway
* Guard against nil tlsContext
* add ingress gateway w/ TLS tests for gRPC and HTTP2
* getAlpnProtocols: add TCP protocol test
* add tests for ingress gateway with grpc/http2 and per-listener TLS config
* add tests for ingress gateway with grpc/http2 and per-listener TLS config
* add Gateway level TLS config with mixed protocol listeners to validate ALPN
* update changelog to include ingress-gateway
* add http/1.1 to http2 ALPN
* go fmt
* fix test on custom-trace-listener
* updating to serf v0.10.1 and memberlist v0.5.0 to get memberlist size metrics and memberlist broadcast queue depth metric
* update changelog
* update changelog
* correcting changelog
* adding "QueueCheckInterval" for memberlist to test
* updating integration test containers to grab latest api
* feat(ingress gateway: support configuring limits in ingress-gateway config entry
- a new Defaults field with max_connections, max_pending_connections, max_requests
is added to ingress gateway config entry
- new field max_connections, max_pending_connections, max_requests in
individual services to overwrite the value in Default
- added unit test and integration test
- updated doc
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
Co-authored-by: Jeff Boruszak <104028618+boruszak@users.noreply.github.com>
Co-authored-by: Dan Stough <dan.stough@hashicorp.com>
* Typos
* Test failing
* Convert values <1ms to decimal
* Fix test
* Update docs and test error msg
* Applied suggested changes to test case
* Changelog file and suggested changes
* Update .changelog/12905.txt
Co-authored-by: Chris S. Kim <kisunji92@gmail.com>
* suggested change - start duration with microseconds instead of nanoseconds
* fix error
* suggested change - floats
Co-authored-by: alex <8968914+acpana@users.noreply.github.com>
Co-authored-by: Chris S. Kim <kisunji92@gmail.com>
* Config-entry: Support proxy config in service-defaults
* Update website/content/docs/connect/config-entries/service-defaults.mdx
Co-authored-by: Jeff Boruszak <104028618+boruszak@users.noreply.github.com>
Prior to #13244, connect proxies and gateways could only be configured by an
xDS session served by the local client agent.
In an upcoming release, it will be possible to deploy a Consul service mesh
without client agents. In this model, xDS sessions will be handled by the
servers themselves, which necessitates load-balancing to prevent a single
server from receiving a disproportionate amount of load and becoming
overwhelmed.
This introduces a simple form of load-balancing where Consul will attempt to
achieve an even spread of load (xDS sessions) between all healthy servers.
It does so by implementing a concurrent session limiter (limiter.SessionLimiter)
and adjusting the limit according to autopilot state and proxy service
registrations in the catalog.
If a server is already over capacity (i.e. the session limit is lowered),
Consul will begin draining sessions to rebalance the load. This will result
in the client receiving a `RESOURCE_EXHAUSTED` status code. It is the client's
responsibility to observe this response and reconnect to a different server.
Users of the gRPC client connection brokered by the
consul-server-connection-manager library will get this for free.
The rate at which Consul will drain sessions to rebalance load is scaled
dynamically based on the number of proxies in the catalog.
http.Transport keeps a pool of connections and should be reused when possible. We instantiate a new http.DefaultTransport for every metrics request, making large numbers of concurrent requests inefficiently spin up new connections instead of reusing open ones.
When a sidecar proxy is registered, a check is automatically added.
Previously, the address this check used was the underlying service's
address instead of the proxy's address, even though the check is testing
if the proxy is up.
This worked in most cases because the proxy ran on the same IP as the
underlying service but it's not guaranteed and so the proper default
address should be the proxy's address.
* draft commit
* add changelog, update test
* remove extra param
* fix test
* update type to account for nil value
* add test for custom passive health check
* update comments and tests
* update description in docs
* fix missing commas
* validate args before deleting proxy defaults
* add changelog
* validate name when normalizing proxy defaults
* add test for proxyConfigEntry
* add comments
`QueryDatacenterOptions` was renamed to `QueryFailoverOptions` without creating
an alias. This adds `QueryDatacenterOptions` back as an alias to
`QueryFailoverOptions` and marks it is deprecated.
Dhia Ayachi
Derek Menteer
Chris Thain
Freddy
Eric Haberkorn
John Maguire
Michael Wilkerson
Luke Kysow
Nitya Dhanushkodi
Melisa Griffin
Andrew Stucki
Valeriia Ruban
Chris S. Kim
Ashvitha
Tyler Wendlandt
John Eikenberry
Ronald
R.B. Boyer
cskh
Nathan Coleman
malizz
Dan Stough
Curt Bushko
Valeriia Ruban
skpratt
Kyle Havlovitz
Paul Banks
Paul Glass
Ashwin Venkatesh
Dan Upton
Frank DiRocco
Nick Irvine
Florian Apolloner
Pier-Luc Caron St-Pierre
Kyle Schochenmaier
John Murret
Jared Kirschner
Dao Thanh Tung
James Oulman
Alexander Scheel
Evan Culver
Iryna Shustava
Tyler Wendlandt
freddygv
Derek Menteer
Michael Klein
Hans Hasselberg
Riddhi Shah
Matt Keeler
Chris Chapman
John Cowen
Nick Ethier
Jared Kirschner
Gabriel Santos
Daniel Graña
Alessandro De Blasis
Jorge Marey
Jorge Marey
Austin Workman