mirror of
https://github.com/apache/impala.git
synced 2026-02-01 12:00:22 -05:00
f8c8fa5b45
testdata/bin/kill-hbase.sh currently uses the generic kill-java-service.sh script to kill the region servers, then the master, and then the zookeeper. Recent versions of HBase become unusable after performing this type of shutdown. The master seems to get stuck trying to recover, even after restarting the minicluster. The root cause in HBase is unclear, but HBase provides the stop-hbase.sh script, which does a more graceful shutdown. This switches tesdata/bin/kill-hbase.sh to use this script, which avoids the recovery problems. Testing: - Ran the test-with-docker.py tests (which does a minicluster restart). Before the change, the HBase tests timed out due to HBase getting stuck recovering. After the change, tests ran normally. - Added a minicluster restart after dataload so that this is tested. Change-Id: I67283f9098c73c849023af8bfa7af62308bf3ed3 Reviewed-on: http://gerrit.cloudera.org:8080/14697 Reviewed-by: Vihang Karajgaonkar <vihang@cloudera.com> Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>