sujeito-operator commented on PR #12731:
URL: https://github.com/apache/gravitino/pull/12731#issuecomment-5480515774
The one non-green check here, `PythonIT (linux/amd64, 17)`, is not reachable
from this diff. Three things, all checkable:
**1. No `pull_request` workflow builds this file.** `docker-image.yml` is
the only workflow that builds `gravitino-ci:kerberos-hive`, and it is
`workflow_dispatch:` only. Nothing in `python-integration-test.yml` builds an
image out of `dev/docker/`.
**2. The integration tests consume pinned published images, not the tree.**
`build.gradle.kts:214-215`:
```kotlin
param.environment("GRAVITINO_CI_HIVE_DOCKER_IMAGE",
"apache/gravitino-ci:hive-0.1.20")
param.environment("GRAVITINO_CI_KERBEROS_HIVE_DOCKER_IMAGE",
"apache/gravitino-ci:kerberos-hive-0.1.6")
```
So `dev/docker/kerberos-hive/Dockerfile` is not an input to any job that
runs here — which is exactly why the follow-up you described (republish the
image, then bump the version in `build.gradle.kts`) is the step that actually
puts this change in front of CI.
**3. That job did not use the kerberos image in any case.**
`HiveContainer.Builder` starts at `DEFAULT_IMAGE` and only swaps to
`KERBEROS_IMAGE` when `kerberosEnabled`, so this run exercised `hive-0.1.20` —
a different Dockerfile from the one edited here.
**What actually failed.** Inside the published hive image the readiness
probe's `hive -e 'show databases;'` returned
```
FAILED: SemanticException org.apache.hadoop.hive.ql.metadata.HiveException:
java.lang.RuntimeException: Unable to instantiate
org.apache.hadoop.hive.ql.metadata.SessionHiveMetaStoreClient
```
on all six attempts between 13:53:46Z and 14:00:09Z, while `hdfs dfsadmin
-report` reported `HDFS is ready` on each one. The container was then logged as
started (`metastore URI: thrift://10.20.31.18:9083`) and the suite carried on;
the two errors are `setUpClass (test_relational_table.TestRelationalTable)` and
`test_catalog_tag_operations (test_supports_tags.TestSupportsTags)`, both
`Connection refused` to that address. HDFS came up; the metastore did not.
Happy to rebase onto current `main` if you would like a fresh run before
merging.
One aside, separate from this PR and only if you agree it is worth an issue:
the probe checks HDFS and the metastore separately, but the run proceeds after
the metastore check has failed every attempt. That turns "the metastore never
started" into two `Connection refused` test errors much further down the log.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
