fix: improve CI resilience for gradle wrapper download and DataNode#88
Draft
MisterRaindrop wants to merge 13 commits intoapache:mainfrom
Draft
fix: improve CI resilience for gradle wrapper download and DataNode#88MisterRaindrop wants to merge 13 commits intoapache:mainfrom
MisterRaindrop wants to merge 13 commits intoapache:mainfrom
Conversation
4e4abda to
e5bf595
Compare
ostinru
reviewed
Mar 27, 2026
| ENV HBASE_URL="https://$APACHE_MIRROR/hbase/$HBASE_VERSION/hbase-$HBASE_VERSION-bin.tar.gz" | ||
| ENV TEZ_URL="https://$APACHE_MIRROR/tez/$TEZ_VERSION/apache-tez-$TEZ_VERSION-bin.tar.gz" | ||
| # Mirror list: try fast mirrors first, fall back to official archive | ||
| ENV APACHE_MIRRORS="repo.huaweicloud.com/apache archive.apache.org/dist" |
Collaborator
There was a problem hiding this comment.
I think that we can cache singlecluster image in github action's cache for 7 days... And we will not see this issue often.
- Set TZ=UTC and -Duser.timezone=UTC for PXF JVM to ensure consistent Parquet INT96 timestamp conversion (ZoneId.systemDefault() in ParquetTypeConverter.java returns OS timezone which differs on Rocky 9) - Pre-cleanup stale Hadoop processes before start-gphd.sh to prevent DataNode BindException on port 50020 - Improve wait_for_hbase() with port 16020 check and 5s stabilization wait instead of simple pgrep (RegionServer can crash after startup) - Add retry logic to HBase RegionServer check in health_check()
The singlecluster configures hbase.regionserver.port=6002<nodeid>, so node 0 listens on port 60020, not the HBase default 16020. Also increase the port wait timeout from 30s to 60s.
The /dev/tcp/localhost/60020 check failed in Docker containers because HBase RegionServer binds to the container IP, not localhost. Revert to simple pgrep + 10s stabilization sleep. Make HBase startup non-fatal so test groups that don't need HBase can still run. Also simplify DataNode pre-cleanup: only kill if stale processes exist.
…e restart - Use fuser -k to force-release DataNode ports (50010/50020/50075/50080) before start-gphd.sh, preventing BindException on CI runners - Fix wait_for_datanode() restart: replace ss|grep pipeline (crashed by set -euo pipefail when grep found no match) with fuser -k - Remove duplicate DataNode start call in restart path - Make HBase/DataNode health checks non-fatal (warn instead of die) so test groups that don't need HBase are not blocked
Add RetryAnalyzer (1 retry) + RetryListener (IAnnotationTransformer) to automatically retry failed tests once. Handles transient failures like HDFS multi-block write timeouts on resource-constrained CI runners.
…RetryAnalyzer not Class
IAnnotationTransformer cannot be registered via @listeners annotation (TestNG limitation - it must be applied before annotations are read). Move registration to maven-surefire-plugin <listener> property.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
fix ci build