Build scan: https://gradle-enterprise.elastic.co/s/svde5ts33nfki
Repro line:
./gradlew ':x-pack:qa:rolling-upgrade:v7.6.2#upgradedClusterTest' --tests "org.elasticsearch.upgrades.UpgradeClusterClientYamlTestSuiteIT.test {p0=upgraded_cluster/30_ml_jobs_crud/Test open old jobs}" \
-Dtests.seed=95C888B6E2499870 \
-Dtests.security.manager=true \
-Dtests.locale=ar-EG \
-Dtests.timezone=Pacific/Gambier \
-Druntime.java=8 \
-Dtests.rest.suite=upgraded_cluster
Reproduces locally?: No
Applicable branches: 7.8 (upgraded from 7.6)
Failure history:
https://build-stats.elastic.co/app/kibana#/discover?_g=(refreshInterval:(pause:!t,value:0),time:(from:'2020-03-03T13:01:09.914Z',mode:absolute,to:'2020-07-03T12:01:15.600Z'))&_a=(columns:!(_source),index:b646ed00-7efc-11e8-bf69-63c8ef516157,interval:auto,query:(language:lucene,query:'%22Test%20open%20old%20jobs%22'),sort:!(process.time-start,desc))
There are many failures of this test over the years, but it has failed for many completely different reasons. The earliest I could find with this problem was on 5th March 2020.
Failure excerpt:
java.lang.AssertionError: Failure at [upgraded_cluster/30_ml_jobs_crud:110]: field [] doesn't have a true value
Expected: not a string equal to "false" ignoring case |
but: was "false"
The relevant bit of the YAML file is:
- do:
indices.exists_alias:
name: ".ml-state-write"
- is_true: ''
We have seen the problem of .ml-state-write being a concrete index rather than an alias in clusters outside of CI. It seems that, very occasionally, it can happen in CI too.
Build scan: https://gradle-enterprise.elastic.co/s/svde5ts33nfki
Repro line:
Reproduces locally?: No
Applicable branches: 7.8 (upgraded from 7.6)
Failure history:
https://build-stats.elastic.co/app/kibana#/discover?_g=(refreshInterval:(pause:!t,value:0),time:(from:'2020-03-03T13:01:09.914Z',mode:absolute,to:'2020-07-03T12:01:15.600Z'))&_a=(columns:!(_source),index:b646ed00-7efc-11e8-bf69-63c8ef516157,interval:auto,query:(language:lucene,query:'%22Test%20open%20old%20jobs%22'),sort:!(process.time-start,desc))
There are many failures of this test over the years, but it has failed for many completely different reasons. The earliest I could find with this problem was on 5th March 2020.
Failure excerpt:
The relevant bit of the YAML file is:
We have seen the problem of
.ml-state-writebeing a concrete index rather than an alias in clusters outside of CI. It seems that, very occasionally, it can happen in CI too.