Known Issues in Apache Solr

Learn about the known issues in Apache Solr, the impact or changes to the functionality, and the workaround.

Known issues identified in Cloudera Runtime 7.3.2.20000 SP2

There are no new known issues in this release.

Known issues identified in Cloudera Runtime 7.3.2.10000 SP1

There are no new known issues in this release.

Known issues identified in Cloudera Runtime 7.3.2.100 CHF 1

There are no new known issues in this release.

Known issues identified in Cloudera Runtime 7.3.2

CDPD-125501: Solr server log files are empty after upgrading Cloudera Manager to 7.13.3

After upgrading to Cloudera Manager 7.13.3.0 on runtimes without the CDPD-125501 launcher fix, the Solr and Infra Solr log files are 0 bytes. The cause is that Cloudera Manager adds the EventCatcher Log4j 2 appender when the Enable Log Event Capture property is on, and the runtime cannot load it.

This issue shows up on the first restart after the upgrade. Solr health checks still pass, and diagnostic bundles contain empty logs.

You have the following options to fix this issue:

  • Recommended: Turn off the Enable Log Event Capture (catch_events) property for Solr and Infra Solr, and restart the service. Note that event capture is already non-functional on affected clusters.
  • Alternative: Set appenders and rootLogger.appenderRefs without EventCatcher in the Solr Server Log4j 2 Advanced Configuration Snippet (Safety Valve). Note that this workaround is brittle across Cloudera Manager updates.

To permanently resolve the cause of this issue, upgrade the runtime to version 7.3.3.0 or later.

CDPD-125782: HBase batch indexer and MapReduceIndexerTool jobs fail with GraalVmProcessor not found

When you run an HBase batch indexer or MapReduceIndexerTool job on Cloudera Runtime 7.3.2 with Cloudera Manager 7.13.3.0 builds up to -31, the job can fail with Provider org.apache.logging.log4j.core.config.plugins.processor.GraalVmProcessor not found. The failure occurs when the morphline contains a java { } block.

The Cloudera Manager event-publish-7.13.3-shaded.jar contains a service entry for a processor that is not available after the jar is shaded. The Cloudera Manager agent places this jar on the MapReduce task classpath. NRT indexing and Cloudera Runtime 7.3.3 are not affected.

You have the following options to fix this issue:

  • Add the Log4j 2.25 jars to the job with the -libjars option:
    -libjars /opt/cloudera/cm/lib/log4j-core-2.25.5.jar,/opt/cloudera/cm/lib/log4j-api-2.25.5.jar
  • Alternatively, append /opt/cloudera/cm/lib/log4j-core-2.25.5.jar and /opt/cloudera/cm/lib/log4j-api-2.25.5.jar to the mapreduce.application.classpath property.
  • For a permanent fix, upgrade Cloudera Manager to 7.13.3.0 build 32 or later, and then run Install YARN MapReduce Framework JARs.

Known issues identified before Cloudera Runtime 7.3.2

Creating Solr collection snapshots using solrctl fails on TLS-enabled clusters
7.3.2 and its SPs and CHFs, 7.3.1 and its SPs and CHFs, 7.1.9 and its SPs and CHFs
When you run the solrctl command to create a collection snapshot (such as solrctl collection --create-snapshot) on a cluster with TLS and Kerberos enabled, the command fails with an IOException occurred when talking to server at: error message.

This issue occurs because the TLS truststore location and password are not exported in the shell environment, preventing the solrctl client from establishing a secure connection to the Apache Solr server over HTTPS. Using the Solr REST API directly to create or list snapshots is not affected.

To resolve this issue, export the truststore path and password in the JAVA_TOOL_OPTIONS environment variable before you run the solrctl command:
JAVA_TOOL_OPTIONS
=
"-Djavax.net.ssl.trustStore=[***PATH_TO_TRUSTSTORE_FILE***] -Djavax.net.ssl.trustStorePassword=[***TRUSTSTORE_PASSWORD***]"

After setting the environment variable, run the solrctl command:

solrctl --jaas
[
***PATH_TO_JAAS_CONF***
]
collection --create-snapshot
[
***SNAPSHOT_NAME***
]
-c
[
***COLLECTION_NAME***
]