[Stack Monitoring] Add stack_monitoring suffix to metrics-* index pattern#137904
Conversation
crespocarlos
changed the title
crespocarlos
changed the title
crespocarlos
added
Team:Infra Monitoring UI - DEPRECATED
|
Pinging @elastic/infra-monitoring-ui (Team:Infra Monitoring UI) |
|
Looks good. Just a note that until #123508 is complete, there are still a few places that won't be updated yet. |
|
@elasticmachine merge upstream |
|
buildkite test this |
|
@elasticmachine merge upstream |
|
Looks like just a docker blip on the last run so re-updating |
|
Should we block this PR until mappings alignment is complete (only logstash left elastic/integrations#3993) ? We can then update the package's data streams to write to the new |
|
When updating a data stream's dataset, not only the index name changes but also the |
…-ref HEAD~1..HEAD --fix'
|
@elasticmachine merge upstream |
| }: CommonIndexPatternArgs & { type?: string }): string { | ||
| const datasetsPattern = `${moduleType ?? '*'}.${dataset ?? '*'}`; | ||
| const datasetsPattern = `${moduleType ?? '*'}.${ | ||
| type === DS_INDEX_PATTERN_METRICS ? 'stack_monitoring.' : '' |
There was a problem hiding this comment.
Added this condition when resolving conflicts and looks like this function is doing too much. We could split it in getMetricsIndexPatterns and getLogsIndexPatterns. the split itself is reasonable and makes more sense with the dataset branching now
There was a problem hiding this comment.
thanks @klacabane. I'll try to improve this.
|
@elasticmachine merge upstream |
|
Pretty sure we're still holding this PR, so I added the blocked label to make sure I'm reviewing things in a good order :) |
|
When testing elastic/integrations#4018 I found that this branch produces a "ghost" standalone cluster when logstash connection errors are present.
Could be that's out of scope for this PR, but it'd be a little weird to ship kibana with this behavior (assuming we haven't already). Update: elastic/integrations#4011 is open to fix that so we probably can skip it for this PR |
|
The ghost standalone will be fixed with #140102 and elastic/integrations#4147 |
|
Should we wait those fixes before merging this one? |
💚 Build Succeeded
Metrics [docs]
History
To update your PR or re-run it, just comment with: |
|
I'm inclined to merge this now since the corresponding integration PR was merged, and the standalone issue does not impact metricbeat/legacy collection mode. |
matschaffer
left a comment
matschaffer
left a comment
There was a problem hiding this comment.
Agreed. I've been using this to test the standalone changes and it seems great.
7 participants
Summary
This PR closes #135382 by updating
metrics-*index pattern - which are generated by the integration packages -, addingstack_monitoringto it, so that the UI can reflect this update on the packagesIt prepares the SM server code for the index pattern update that will happen on the SM integration packages
For maintainers
There is actually a limit of 255 characters for index names for index name, but I don't expect that adding this suffix will cause problems in the future. Considering the following example:
.ds-metrics-elasticsearch.index_recovery-default-2022.08.02-000001= 66 charactersThis PR can only be merged once the integration packages indices have been updated