Add leader election for autodiscover #20281
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: chrismark <[email protected]>
ChrsMark
added
enhancement
in progress
botelastic
bot
added
needs_team
❕ Build Aborted
Expand to view the summary
Build stats
Test stats 🧪
Log outputExpand to view the last 100 lines of log output
|
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
|
Tested this on a 3-node cluster on GKE. Seems to work:
I will open it for review by now. |
|
Pinging @elastic/integrations-platforms (Team:Platforms) |
Signed-off-by: chrismark <[email protected]>
jsoriano
reviewed
Aug 3, 2020
Signed-off-by: chrismark <[email protected]>
ChrsMark
changed the title
jsoriano
reviewed
Aug 3, 2020
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
|
Thanks for the review and the (offline) suggestions @jsoriano! |
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
jsoriano
added
needs_backport
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
Signed-off-by: chrismark <[email protected]>
jsoriano
reviewed
Aug 6, 2020
|
|
||
| // Start for EventManager interface. | ||
| func (p *leaderElectionManager) Start() { | ||
| ctx, cancel := context.WithCancel(context.Background()) |
There was a problem hiding this comment.
Nit.
Suggested change
| ctx, cancel := context.WithCancel(context.Background()) | |
| ctx, cancel := context.WithCancel(context.TODO()) |
| func (p *leaderElectionManager) startLeaderElector(ctx context.Context, lec leaderelection.LeaderElectionConfig) { | ||
| le, err := leaderelection.NewLeaderElector(lec) | ||
| if err != nil { | ||
| p.logger.Errorf("leader election lock GAINED, id %v", err) |
There was a problem hiding this comment.
This doesn't look like an error 🙂 Log it at the info or debug level?
There was a problem hiding this comment.
This actually an error..but with wrong error message. This piece is "copied" from RunOrDie (https://github.com/kubernetes/client-go/blob/36233866f1c7c0ad3bdac1fc466cb5de3746cfa2/tools/leaderelection/leaderelection.go#L214). See the watchdog related comment for more info.
| if err != nil { | ||
| p.logger.Errorf("leader election lock GAINED, id %v", err) | ||
| } | ||
| if lec.WatchDog != nil { |
There was a problem hiding this comment.
What is this watchdog? we are never setting it.
There was a problem hiding this comment.
This was mostly because I tried to "copy" the RunOrDie of https://github.com/kubernetes/client-go/blob/36233866f1c7c0ad3bdac1fc466cb5de3746cfa2/tools/leaderelection/leaderelection.go#L219, because of an async problem described at kubernetes/client-go#837 .
Good catch I will remove it, it's redundant.
CHANGELOG.next.asciidoc
Outdated
| @@ -367,6 +367,7 @@ https://github.com/elastic/beats/compare/v7.0.0-alpha2...master[Check the HEAD d | |||
| - Added the `max_cached_sessions` option to the script processor. {pull}19562[19562] | |||
| - Add support for DNS over TLS for the dns_processor. {pull}19321[19321] | |||
| - Set index.max_docvalue_fields_search in index template to increase value to 200 fields. {issue}20215[20215] | |||
| - Add leader election for autodiscover. {pull}20281[20281] | |||
There was a problem hiding this comment.
Suggested change
| - Add leader election for autodiscover. {pull}20281[20281] | |
| - Add leader election for Kubernetes autodiscover. {pull}20281[20281] |
Signed-off-by: chrismark <[email protected]>
|
jenkins run the tests please |
jsoriano
approved these changes
Aug 8, 2020
💔 Tests Failed
Expand to view the summary
Build stats
Test stats 🧪
Test errorsExpand to view the tests failures
Steps errorsExpand to view the steps failures
Log outputExpand to view the last 100 lines of log output
|
|
@jsoriano I think the only issue with the tests is https://travis-ci.org/github/elastic/beats/jobs/716456468#L1590 which is also failing in general currently. I think we can proceed with merging, wdyt? |
Yes, I think the issue in CI is not related. |
ChrsMark
added a commit
to ChrsMark/beats
that referenced
this pull request
Aug 10, 2020
(cherry picked from commit 9ab9b97)
6 tasks
ChrsMark
removed
the
needs_backport
7 tasks
ChrsMark
added a commit
that referenced
this pull request
Aug 11, 2020
7 tasks
melchiormoulin
pushed a commit
to melchiormoulin/beats
that referenced
this pull request
Oct 14, 2020
jorgelbg
added a commit
to jorgelbg/beats
that referenced
this pull request
Nov 16, 2020
After PR elastic#20281 was merged, the signature of the `autodiscover.ProviderBuilder` changed to accept an additional name (string) parameter.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Implements leader election as part of kubernetes autodiscover provider. With this, when the beat is elected as leader it will start cluster scope metricsets.
Under the hood, all leader candidates will try to gain the lock on Lease object (https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.18/#lease-v1-coordination-k8s-io). When the lock is gained then the configured cluster wide mericsets will get started.
Why is it important?
To get rid of
Deploymentprocedure for the singleton Metricbeat instance which is needed for the cluster scope metricsets.Sample configuration that will start
state_nodemetricset when a leadership is gained:Then we can monitor the
leaseto check which provider holds the lock:Checklist
CHANGELOG.next.asciidocorCHANGELOG-developer.next.asciidoc.How to test this PR locally
It is recommended to testing this along with #20512 for easiness of configuration/setup etc
Make sure that events for
state_nodeare being collected from only one metricbeat instance. You can the in the logs and verify that only one metricbeat has gained the lock so far.Monitor the
leaseobject to keep track of the lock holder:watch kubectl describe lease beats-cluster-leaderStop the Metricbeat instance that currently holds the lock and make sure that another one takes over and we keep getting events from
state-node. Make sure that theleasechanged holder (logs will also let us know that a new metricset is getting started)Also test without setting
identifierand check that it takes the default value properly likemetricbeat-cluster-leader.Check with old autodiscover (template based/hints based) for possible regressions.
Related issues