IMPALA-11990: Make actual failures clearer

A hack to cleanup after Hbase fails when services haven't been started
yet (which is always at least once in a CI run) with a large error
message. That error isn't useful and can be misleading for people
reviewing test logs. Suppress it.

Guards data load for Ozone as a usable snapshot is required. Also fixes
a typo in fixed issues.

Change-Id: Idc37d03780fca35427b977524b2b97a6892c88f7
Reviewed-on: http://gerrit.cloudera.org:8080/19459
Reviewed-by: Gabor Kaszab <gaborkaszab@cloudera.com>
Tested-by: Michael Smith <michael.smith@cloudera.com>
This commit is contained in:
Michael Smith
2023-01-30 10:44:30 -08:00
parent 490dd7b115
commit 39fea06f2b
3 changed files with 6 additions and 5 deletions

View File

@@ -150,7 +150,7 @@ under the License.
<li><xref href="https://issues.apache.org/jira/browse/IMPALA-8061"
format="html" scope="external">IMPALA-8061</xref> - Impala correctly
initializes<codeph> S3_ACCESS_VALIDATED</codeph> variable to zero
when <codeph>TARGET_FILESYSTEM=3</codeph>.</li>
when <codeph>TARGET_FILESYSTEM=s3</codeph>.</li>
<li><xref href="https://issues.apache.org/jira/browse/IMPALA-8154"
format="html" scope="external">IMPALA-8154</xref> - Disabled the
Kerberos<codeph> auth_to_local</codeph> setting to prevent

View File

@@ -136,10 +136,10 @@ elif [ $SKIP_SNAPSHOT_LOAD -eq 0 ]; then
if [[ "${TARGET_FILESYSTEM}" == "isilon" || "${TARGET_FILESYSTEM}" == "s3" || \
"${TARGET_FILESYSTEM}" == "local" || "${TARGET_FILESYSTEM}" == "gs" || \
"${TARGET_FILESYSTEM}" == "cosn" || "${TARGET_FILESYSTEM}" == "oss" || \
"${TARGET_FILESYSTEM}" == "obs" ]] ; then
"${TARGET_FILESYSTEM}" == "obs" || "${TARGET_FILESYSTEM}" == "ozone" ]] ; then
echo "ERROR in $0 at line $LINENO: A schema change has been detected in the"
echo "metadata, but it cannot be loaded on isilon, s3, gcs, cos, oss or local"
echo "and the target file system is ${TARGET_FILESYSTEM}. Exiting."
echo "metadata, but it cannot be loaded on isilon, s3, gcs, cos, oss, obs, ozone,"
echo "or local and the target file system is ${TARGET_FILESYSTEM}. Exiting."
# Generate an explicit JUnitXML symptom report here for easier triaging
${IMPALA_HOME}/bin/generate_junitxml.py --phase=dataload \
--step=check-schema-diff.sh --error "${SCHEMA_MISMATCH_ERROR}"

View File

@@ -38,4 +38,5 @@ rm -rf /tmp/hbase-*
# HACK: Some jobs have seen the HBase master fail to initialize with mesages like:
# "Master startup cannot progress, in holding-pattern until region onlined."
# Anecdotally, removing the MasterProcWALs directory avoids the issue.
hdfs dfs -Dozone.client.failover.max.attempts=3 -rm /hbase/MasterProcWALs/* || true
hdfs dfs -Dozone.client.failover.max.attempts=3 -rm /hbase/MasterProcWALs/* \
>& /dev/null || true