Starting from and extending the mechanism introduced in #12110 we can specially handle the 3 main special Consul RPC endpoints that react to many config entries in a single blocking query in Connect:
- `DiscoveryChain.Get`
- `ConfigEntry.ResolveServiceConfig`
- `Intentions.Match`
All of these will internally watch for many config entries, and at least one of those will likely be not found in any given query. Because these are blends of multiple reads the exact solution from #12110 isn't perfectly aligned, but we can tweak the approach slightly and regain the utility of that mechanism.
### No Config Entries Found
In this case, despite looking for many config entries none may be found at all. Unlike #12110 in this scenario we do not return an empty reply to the caller, but instead synthesize a struct from default values to return. This can be handled nearly identically to #12110 with the first 1-2 replies being non-empty payloads followed by the standard spurious wakeup suppression mechanism from #12110.
### No Change Since Last Wakeup
Once a blocking query loop on the server has completed and slept at least once, there is a further optimization we can make here to detect if any of the config entries that were present at specific versions for the prior execution of the loop are identical for the loop we just woke up for. In that scenario we can return a slightly different internal sentinel error and basically externally handle it similar to #12110.
This would mean that even if 20 discovery chain read RPC handling goroutines wakeup due to the creation of an unrelated config entry, the only ones that will terminate and reply with a blob of data are those that genuinely have new data to report.
### Extra Endpoints
Since this pattern is pretty reusable, other key config-entry-adjacent endpoints used by `agent/proxycfg` also were updated:
- `ConfigEntry.List`
- `Internal.IntentionUpstreams` (tproxy)
Many places in consul already treated node names case insensitively.
The state store indexes already do it, but there are a few places that
did a direct byte comparison which have now been corrected.
One place of particular consideration is ensureCheckIfNodeMatches
which is executed during snapshot restore (among other places). If a
node check used a slightly different casing than the casing of the node
during register then the snapshot restore here would deterministically
fail. This has been fixed.
Primary approach:
git grep -i "node.*[!=]=.*node" -- ':!*_test.go' ':!docs'
git grep -i '\[[^]]*member[^]]*\]
git grep -i '\[[^]]*\(member\|name\|node\)[^]]*\]' -- ':!*_test.go' ':!website' ':!ui' ':!agent/proxycfg/testing.go:' ':!*.md'
There are some cross-config-entry relationships that are enforced during
"graph validation" at persistence time that are required to be
maintained. This means that config entries may form a digraph at times.
Config entry replication procedes in a particular sorted order by kind
and name.
Occasionally there are some fixups to these digraphs that end up
replicating in the wrong order and replicating the leaves
(ingress-gateway) before the roots (service-defaults) leading to
replication halting due to a graph validation error related to things
like mismatched service protocol requirements.
This PR changes replication to give each computed change (upsert/delete)
a fair shot at being applied before deciding to terminate that round of
replication in error. In the case where we've simply tried to do the
operations in the wrong order at least ONE of the outstanding requests
will complete in the right order, leading the subsequent round to have
fewer operations to do, with a smaller likelihood of graph validation
errors.
This does not address all scenarios, but for scenarios where the edits
are being applied in the wrong order this should avoid replication
halting.
Fixes#9319
The scenario that is NOT ADDRESSED by this PR is as follows:
1. create: service-defaults: name=new-web, protocol=http
2. create: service-defaults: name=old-web, protocol=http
3. create: service-resolver: name=old-web, redirect-to=new-web
4. delete: service-resolver: name=old-web
5. update: service-defaults: name=old-web, protocol=grpc
6. update: service-defaults: name=new-web, protocol=grpc
7. create: service-resolver: name=old-web, redirect-to=new-web
If you shutdown dc2 just before (4) and turn it back on after (7)
replication is impossible as there is no single edit you can make to
make forward progress.
* Add %panel CSS component
* Deprecate old menu-panel component
* Various smallish tweaks to disclosure-menu
* Move all menus in the app chrome to use new DisclosureMenu
* Follow up CSS to move all app chrome menus to new components
* Don't prevent default any events from anchors
* Add a tick to click steps
* Delete collapsible notices component and related helper
* Add relative t action/helper to our Route component
* Replace single use CollapsibleNotices with multi-use Disclosure
* Parse datacenter from request
- Parse the value of the datacenter from the create/delete requests for AuthMethods and BindingRules so that they can be created in and deleted from the datacenters specified in the request.
By using the query results as state.
Blocking queries are efficient when the query matches some results,
because the ModifyIndex of those results, returned as queryMeta.Mindex,
will never change unless the items themselves change.
Blocking queries for non-existent items are not efficient because the
queryMeta.Index can (and often does) change when other entities are
written.
This commit reduces the churn of these queries by using a different
comparison for "has changed". Instead of using the modified index, we
use the existence of the results. If the previous result was "not found"
and the new result is still "not found", we know we can ignore the
modified index and continue to block.
This is done by setting the minQueryIndex to the returned
queryMeta.Index, which prevents the query from returning before a state
change is observed.
We've noticed that a trace that is captured over the full duration is
too large to open on most machines. A trace.out captured over just the
interval period (30s by default) should be a more than enough time to
capture trace data.
This will both save on unnecessary raft operations as well as
unnecessarily incrementing the raft modify index of config entries
subject to no-op updates.
The race detector noticed this initially in `TestAgentConfigWatcherSidecarProxy` but it is not restricted to just tests.
The two main changes here were:
- ensure that before we mutate the internal `agent/local` representation of a Service (for tags or VIPs) we clone those fields
- ensure that there's no function argument joint ownership between the caller of a function and the local state when calling `AddService`, `AddCheck`, and related using `copystructure` for now.
* First phase of refactoring PermissionDeniedError
Add extended type PermissionDeniedByACLError that captures information
about the accessor, particular permission type and the object and name
of the thing being checked.
It may be worth folding the test and error return into a single helper
function, that can happen at a later date.
Signed-off-by: Mark Anderson <manderson@hashicorp.com>
This commit excludes the health of any service instances from the Node Listing page. This means that if you are viewing the Node listing page you will only see failing nodes if there are any Node Checks failing, Service Instance Health checks are no longer taken into account.
Co-authored-by: Jamie White <jamie@jgwhite.co.uk>
We noticed that the Service Instance listing on both Node and Service views where not taking into account proxy instance health. This fixes that up so that the small health check information in each Service Instance row includes the proxy instances health checks when displaying Service Instance health (afterall if the proxy instance is unhealthy then so is the service instance that it should be proxying)
* Refactor Consul::InstanceChecks with docs
* Add to-hash helper, which will return an object keyed by a prop
* Stop using/relying on ember-data type things, just use a hash lookup
* For the moment add an equivalent "just give me proxies" model prop
* Start stitching things together, this one requires an extra HTTP request
..previously we weren't even requesting proxies instances here
* Finish up the stitching
* Document Consul::ServiceInstance::List while I'm here
* Fix up navigation mocks Name > Service
Fixes#11876
This enforces that multiple xDS mutations are not issued on the same ADS connection at once, so that we can 100% control the order that they are applied. The original code made assumptions about the way multiple in-flight mutations were applied on the Envoy side that was incorrect.
When a wildcard xDS type (LDS/CDS/SRDS) reconnects from a delta xDS stream,
prior to envoy `1.19.0` it would populate the `ResourceNamesSubscribe` field
with the full list of currently subscribed items, instead of simply omitting it
to infer that it wanted everything (which is what wildcard mode means).
This upstream issue was filed in envoyproxy/envoy#16063 and fixed in
envoyproxy/envoy#16153 which went out in Envoy `1.19.0` and is fixed in later
versions (later refactored in envoyproxy/envoy#16855).
This PR conditionally forces LDS/CDS to be wildcard-only even when the
connected Envoy requests a non-wildcard subscription, but only does so on
versions prior to `1.19.0`, as we should not need to do this on later versions.
This fixes the failure case as described here: #11833 (comment)
Co-authored-by: Huan Wang <fredwanghuan@gmail.com>
- Adding a 'Partition' and 'RetryJoin' command allows test cases where
one would like to spin up a Consul Agent in a non-default partition to
test use-cases that are common when enabling Admin Partition on
Kubernetes.
The fix here is two fold:
- We shouldn't be providing the DataSource (which loads the data) with an id when we are creating from within a folder (in the buggy code we are providing the parentKey of the new KV you are creating)
- Being able to provide an empty id to the DataSource/KV repository and that repository responding with a newly created object is more towards the "new way of doing forms", therefore the corresponding code to return a newly created ember-data object. As we changed the actual bug in point 1 here, we need to make sure the repository responds with an empty object when the request id is empty.
* xds: refactor ingress listener SDS configuration
* xds: update resolveListenerSDS call args in listeners_test
* ingress: add TLS min, max and cipher suites to GatewayTLSConfig
* xds: implement envoyTLSVersions and envoyTLSCipherSuites
* xds: merge TLS config
* xds: configure TLS parameters with ingress TLS context from leaf
* xds: nil check in resolveListenerTLSConfig validation
* xds: nil check in makeTLSParameters* functions
* changelog: add entry for TLS params on ingress config entries
* xds: remove indirection for TLS params in TLSConfig structs
* xds: return tlsContext, nil instead of ambiguous err
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
* xds: switch zero checks to types.TLSVersionUnspecified
* ingress: add validation for ingress config entry TLS params
* ingress: validate listener TLS config
* xds: add basic ingress with TLS params tests
* xds: add ingress listeners mixed TLS min version defaults precedence test
* xds: add more explicit tests for ingress listeners inheriting gateway defaults
* xds: add test for single TLS listener on gateway without TLS defaults
* xds: regen golden files for TLSVersionInvalid zero value, add TLSVersionAuto listener test
* types/tls: change TLSVersion to string
* types/tls: update TLSCipherSuite to string type
* types/tls: implement validation functions for TLSVersion and TLSCipherSuites, make some maps private
* api: add TLS params to GatewayTLSConfig, add tests
* api: add TLSMinVersion to ingress gateway config entry test JSON
* xds: switch to Envoy TLS cipher suite encoding from types package
* xds: fixup validation for TLSv1_3 min version with cipher suites
* add some kitchen sink tests and add a missing struct tag
* xds: check if mergedCfg.TLSVersion is in TLSVersionsWithConfigurableCipherSuites
* xds: update connectTLSEnabled comment
* xds: remove unsued resolveGatewayServiceTLSConfig function
* xds: add makeCommonTLSContextFromLeafWithoutParams
* types/tls: add LessThan comparator function for concrete values
* types/tls: change tlsVersions validation map from string to TLSVersion keys
* types/tls: remove unused envoyTLSCipherSuites
* types/tls: enable chacha20 cipher suites for Consul agent
* types/tls: remove insecure cipher suites from allowed config
TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA256 and TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256 are both explicitly listed as insecure and disabled in the Go source.
Refs https://cs.opensource.google/go/go/+/refs/tags/go1.17.3:src/crypto/tls/cipher_suites.go;l=329-330
* types/tls: add ValidateConsulAgentCipherSuites function, make direct lookup map private
* types/tls: return all unmatched cipher suites in validation errors
* xds: check that Envoy API value matching TLS version is found when building TlsParameters
* types/tls: check that value is found in map before appending to slice in MarshalEnvoyTLSCipherSuiteStrings
* types/tls: cast to string rather than fmt.Printf in TLSCihperSuite.String()
* xds: add TLSVersionUnspecified to list of configurable cipher suites
* structs: update note about config entry warning
* xds: remove TLS min version cipher suite unconfigurable test placeholder
* types/tls: update tests to remove assumption about private map values
Co-authored-by: R.B. Boyer <rb@hashicorp.com>
* Make sure the mocks reflect the requested partition/namespace
* Ensure partition is passed through to the HTTP adapter
* Pass AuthMethod object through to TokenSource in order to use Partition
* Change up docs and add potential improvements for future
* Pass the query partition back onto the response
* Make sure the OIDC callback mock returns a Partition
* Enable OIDC provider mock overwriting during acceptance testing
* Make sure we can enable partitions and SSO post bootup only required
...for now
* Wire up oidc provider mocking
* Add SSO full auth flow acceptance tests
* ui: Don't even ask whether we are authorized for a KV...
...just let the actual API tell us in the response, thin-client style.
* Add some similar commenting for previous PRs related to this problem
* Add some less fake API data
* Rename the models class so as to not be confused with JS Proxies
* Rearrange routlets slightly and add some initial outletFor tests
* Move away from a MeshChecks computed property and just use a helper
* Just use ServiceChecks for healthiness filtering for the moment
* Make TProxy cookie configurable
* Amend exposed paths and upstreams so they know about meta AND proxy
* Slight bit of TaggedAddresses refactor while I was checking for `meta` etc
* Document CONSUL_TPROXY_ENABLE
We recently changed the intentions form to take a full model of a dc rather than just the string identifier (so {Name: 'dc', Primary: true} vs just 'dc' in order to know whether the DC is the primary or not.
Unfortunately, we only did this on the global intentions page not the per service intentions page. This makes it impossible to save an intention from the per service intention page (whilst you can still save intentions from the global intention page as normal).
The fix here pretty much copy/pastes the approach taken in the global intention edit template over to the per service intention edit template.
Tests have been added for creation in the per service intention section, which again are pretty much just copied from the global one, unfortunately this didn't exist previously which would have helped prevent this.
Update the `/agent/check/deregister/` API endpoint to return a 404
HTTP response code when an attempt is made to de-register a check ID
that does not exist on the agent.
This brings the behavior of /agent/check/deregister/ in line with the
behavior of /agent/service/deregister/ which was changed in #10632 to
similarly return a 404 when de-registering non-existent services.
Fixes#5821
* clone the service under lock to avoid a data race
* add change log
* create a struct and copy the pointer to mutate it to avoid a data race
* fix failing test
* revert added space
* add comments, to clarify the data race.
Error messages related to service and check operations previously included
the following substrings:
- service %q
- check %q
From this error message, it isn't clear that the expected field is the ID for
the entity, not the name. For example, if the user has a service named test,
the error message would read 'Unknown service "test"'. This is misleading -
a service with that *name* does exist, but not with that *ID*.
The substrings above have been modified to make it clear that ID is needed,
not name:
- service with ID %q
- check with ID %q
* ui: Add login button to per service intentions for zero results
* Add login button and consistent header for when you have zero nodes
* `services` doesn't exists use `items` consequently:
Previous to this fix we would not show a more tailored message for when
you empty result set was due to a user search rather than an empty
result set straight from the backend
* Fix `error` > `@error` in ErrorState plus code formatting and more docs
* Changelog
When a URL path is not found, return a non-empty message with the 404 status
code to help the user understand what went wrong. If the URL path was not
prefixed with '/v1/', suggest that may be the cause of the problem (which is a
common mistake).
* Update disco fixtures now we have partitions
* Add virtual-admin-6 fixture with partition 'redirects' and failovers
* Properly cope with extra partition segment for splitters and resolvers
* Make 'redirects' and failovers look/act consistently
* Fixup some unit tests
Fixes a bug whereby servers present in multiple network areas would be
properly segmented in the Router, but not in the gRPC mirror. This would
lead servers in the current datacenter leaving from a network area
(possibly during the network area's removal) from deleting their own
records that still exist in the standard WAN area.
The gRPC client stack uses the gRPC server tracker to execute all RPCs,
even those targeting members of the current datacenter (which is unlike
the net/rpc stack which has a bypass mechanism).
This would manifest as a gRPC method call never opening a socket because
it would block forever waiting for the current datacenter's pool of
servers to be non-empty.
types: add TLS constants
types: distinguish between human and Envoy serialization for TLSVersion constants
types: add DeprecatedAgentTLSVersions for backwards compatibility
types: add methods for printing TLSVersion as strings
types: add TLSVersionInvalid error value
types: add a basic test for TLSVersion comparison
types: add TLS cihper suite mapping using IANA constant names and values
types: adding ConsulAutoConfigTLSVersionStrings
changelog: add entry for TLSVersion and TLSCipherSuite types
types: initialize TLSVerison constants starting at zero
types: remove TLSVersionInvalid < 0 test
types: update note for ConsulAutoConfigTLSVersionStrings
types: programmatically invert TLSCipherSuites for HumanTLSCipherSuiteStrings lookup map
Co-authored-by: Dan Upton <daniel@floppy.co>
types: add test for TLSVersion zero-value
types: remove unused EnvoyTLSVersionStrings
types: implement MarshalJSON for TLSVersion
types: implement TLSVersionUnspecified as zero value
types: delegate TLS.MarshalJSON to json.Marshal, use ConsulConfigTLSVersionStrings as default String() values
Co-authored-by: Dan Upton <daniel@floppy.co>
The test added in this commit shows the problem. Previously the
SigningKeyID was set to the RootCert not the local leaf signing cert.
This same bug was fixed in two other places back in 2019, but this last one was
missed.
While fixing this bug I noticed I had the same few lines of code in 3
places, so I extracted a new function for them.
There would be 4 places, but currently the InitializeCA flow sets this
SigningKeyID in a different way, so I've left that alone for now.
We were not adding the local signing cert to the CARoot. This commit
fixes that bug, and also adds support for fixing existing CARoot on
upgrade.
Also update the tests for both primary and secondary to be more strict.
Check the SigningKeyID is correct after initialization and rotation.
This commit uses all our new ways of doing things to Lock Sessions and their interactions with KV and Nodes. This is mostly around are new under-the-hood things, but also I took the opportunity to upgrade some of the CSS to reuse some of our CSS utils that have been made over the past few months (%csv-list and %horizontal-kv-list).
Also added (and worked on existing) documentation for Lock Session related components.
- Moves where they appear up to the <App /> component.
- Instead of a <Notification /> wrapping component to move whatever you use for a notification up to where they need to appear (via ember-cli-flash), we now use a {{notification}} modifier now we have modifiers.
- Global notifications/flashes are no longer special styles of their own. You just use the {{notification}} modifier to hoist whatever component/element you want up to the top of the page. This means we can re-use our existing <Notice /> component for all our global UI notifications (this is the user visible change here)
* Upgrade AuthForm and document current state a little better
* Hoist SSO out of the AuthForm
* Bare minimum admin partitioned SSO
also:
ui: Tabbed Login with Token or SSO interface (#11619)
- I upgraded our super old, almost the first ember component I wrote, to use glimmer/almost template only. This should use slots/contextual components somehow, but thats a bigger upgrade so I didn't go that far.
- I've been wanting to upgrade the shape of our StateChart component for a very long while now, here its very apparent that it would be much better to do this sooner rather than later. I left it as is for now, but there will be a PR coming soon with a slight reshaping of this component.
- Added a did-upsert modifier which is a mix of did-insert/did-update
- Documentation added/amended for all the new things.
R.B. Boyer
Chris S. Kim
Daniele Vazzola
chinmaym07
Evan Culver
Evan Culver
Daniel Nephin
John Cowen
Kyle Havlovitz
Ashwin Venkatesh
Florian Apolloner
Mark Anderson
Freddy
freddygv
Dhia Ayachi
claire labry
Dan Upton
Michele Degges
Mike Morris
Blake Covarrubias
Jared Kirschner
Jared Kirschner
littlestar642
FFMMM
haxandmat
Matt Keeler