-
Notifications
You must be signed in to change notification settings - Fork 24.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
CI: org.elasticsearch.index.shard.IndexShardIT#testIndexCanChangeCustomDataPath #43964
Comments
Pinging @elastic/es-distributed |
I've assigned this to @tlrx as I suspect it might be related to the translog trimming work which now runs on closed indices. The test assumes that the data on disk does not change for a closed index. |
Another more recent failure: https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+7.x+multijob-unix-compatibility/os=ubuntu-18.04&&immutable/119/console Doesn't reproduce for me as well. |
The test IndexShardIT.testIndexCanChangeCustomDataPath() fails on 7.x and 7.3 because the translog cannot be recovered. While I can't reproduce the issue, I think it has been introduced in #43752 which changed ReadOnlyEngine so that it opens the translog in its constructor in order to load the translog stats. This opening writes a new checkpoint file, but because 7.x/7.3 does not wait for shards to be started after being closed, the test immediately starts to copy shard files to a new directory and possibly does not copy all the required translog files. By waiting for the shards to be started after being closed, we ensure that the shards (and engines) have been correctly initialized and that the translog checkpoint file is not currently being written. closes #43964
The test IndexShardIT.testIndexCanChangeCustomDataPath() fails on 7.x and 7.3 because the translog cannot be recovered. While I can't reproduce the issue, I think it has been introduced in #43752 which changed ReadOnlyEngine so that it opens the translog in its constructor in order to load the translog stats. This opening writes a new checkpoint file, but because 7.x/7.3 does not wait for shards to be started after being closed, the test immediately starts to copy shard files to a new directory and possibly does not copy all the required translog files. By waiting for the shards to be started after being closed, we ensure that the shards (and engines) have been correctly initialized and that the translog checkpoint file is not currently being written. closes #43964
Fixed in #43978 |
Example reproduction
Example failure
https://scans.gradle.com/s/zngdfh37brdo2/
Frequency
11 times since yesterday
Will mute this
The text was updated successfully, but these errors were encountered: