BWC test failures for MultiVersionRepositoryAccessIT #50819
Comments
mark-vieira
added
:Distributed Coordination/Snapshot/Restore
|
Pinging @elastic/es-distributed (:Distributed/Snapshot/Restore) |
|
@mark-vieira yea I know how to fix this (already did in |
|
No rush. These builds only run every 8 hours so we are talking maybe 1 or 2 more failures so hardly much in the way of noise in the grand scheme of things. Thanks, Armin! |
We need the same fix we did in `7.x` (elastic#50797) and only get snapshot status for the current version or older. Otherwise these tests break for e.g.`7.0.1` due to the same index metadata incompatibility. Closes elastic#50819
|
@original-brownbear we are getting more failures on the https://elasticsearch-ci.elastic.co/view/Elasticsearch%207.x/job/elastic+elasticsearch+7.x+bwc/483/ https://gradle-enterprise.elastic.co/s/72ecamxlrgnc4/tests/b25xfi2kvu43g-g5ybnrrraxiry |
|
@mark-vieira ah sorry about that! Will fix it right away in my morning in 12h |
|
Thanks, Armin! |
|
For the sake of completeness: We have two more build failures in:
(omitting build scans as they don't add any new information) |
The HLRC doesn't like what its getting back from some older 6.x versions for the restore status so I moved that request to the low level client. Closes elastic#50819
|
Relatively trivial fix incoming in #51006 :) |
|
Thanks again, Armin! |
|
npnp closed via #51006 |
* Fix MultiVersionRepository BwC Tests The HLRC doesn't like what its getting back from some older 6.x versions for the restore status so I moved that request to the low level client. Closes #50819
|
@original-brownbear we are still getting failures on https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+7.x+default-distro+bwc/487/ |
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes elastic#50819
|
Sorry about that, the failure we're seeing there could in fact also hit |
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes #50819
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes elastic#50819
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes elastic#50819
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes #50819
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes #50819
|
Sorry @original-brownbear it looks like it's still failing in 7.x (694b8ab) BWC against 6.4.3 |
|
I haven't checked whether the logs / causes are identical, but it looks like it's broken from [6.3.0, 6.5.4] |
|
Same type of failures also occur for 7.6 BWC tests against from 6.3.0 to 6.6.4 https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+7.6+default-distro+bwc/13/ |
|
Looks like there is yet another random failure left here unrelated to those fixed so far. I'm on it again. |
If the rollup jobs index doesn't exist for some reason (like running against a 6.x cluster) we should just assume the jobs have been cleaned up and move on. Closes elastic#50819
|
Fix incoming in #51246 |
* Fix Rest Tests Failing to Cleanup Rollup Jobs If the rollup jobs index doesn't exist for some reason (like running against a 6.x cluster) we should just assume the jobs have been cleaned up and move on. Closes #50819
* Fix Rest Tests Failing to Cleanup Rollup Jobs If the rollup jobs index doesn't exist for some reason (like running against a 6.x cluster) we should just assume the jobs have been cleaned up and move on. Closes elastic#50819
* Fix Rest Tests Failing to Cleanup Rollup Jobs If the rollup jobs index doesn't exist for some reason (like running against a 6.x cluster) we should just assume the jobs have been cleaned up and move on. Closes elastic#50819
We need the same fix we did in `7.x` (elastic#50797) and only get snapshot status for the current version or older. Otherwise these tests break for e.g.`7.0.1` due to the same index metadata incompatibility. Closes elastic#50819
If some internal `.watcher` or so index gets created during these tests then the shard counts on snapshot restores and creates won't match up with expectations. Fixed by only creating the snapshot for the test index Closes elastic#50819
Our most recent
masterBWC test run failed against versions 7.0.x and 7.1.x with the same test failures indicating we either made a BWC incompatible change inmasteror perhaps just the test needs updating.The build run in question included this commit, which might potentially be related. @original-brownbear any thoughts?
Example build scan: https://gradle-enterprise.elastic.co/s/dpoop7fvn2rwq/tests/failed
Jenkins build with links to each BWC version: https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+master+bwc/471/
The text was updated successfully, but these errors were encountered: