Known Issues in Cloudera Search

Learn about the known issues in Cloudera Search, the impact or changes to the functionality, and the workaround.

Known Issues

TSB 2022-535: Ranger audit retention settings in Solr are not honored
The audits present in the ranger_audits collection in the Solr service of Data Lake do not get deleted based on the retention period set. The default retention period is 90 days.

This is caused by the incorrect order of processors in the configuration (solrconfig.xml) used by the ranger_audits collection.

  • Older audits based on the retention period are not deleted that leads to:

    • An increase in space utilization under the Solr Data Directory. You can check the location being used for the “Solr Data Directory” setting in the Solr service configuration.

    • Increases the memory usage of Solr service

    • Backup-Restore of Solr collection would take longer

Action required
  • Workaround
    • Update Solr configuration and clean old audits

      This is a two-step process. Note that this process must be completed for any Data Lake that has used an affected runtime version, even if the Data Lake has since been upgraded to a newer runtime.

      1. Part 1: Manually correct the configurations to ensure that Solr retention value is enforced for new documents that are added.

      2. Part 2: Manually cleanup documents that are older than 90 days.

  • Upgrade (recommended)
    Upgrade to runtime 7.2.11+

    If the Data Lake is not upgraded, then the manual configuration Solr changes will be overwritten if the Data Lake is repaired or upgraded to runtime lower than 7.2.11. To ensure the changes are persisted, the Data Lake must be upgraded to runtime 7.2.11 or higher. See below for additional upgrade details. Note that the manual steps in Parts 1 and 2 above still need to be done even if the Data Lake is upgraded.

Knowledge article
For the latest update on this issue see the corresponding Knowledge article: TSB 2021-535: Ranger audit retention settings in Solr are not honored.
MapreduceIndexerTool performance problem in CDP

The reduce step of the MorphlineMapper task of the MapReduceIndexerTool (MRIT) can take very long to finish in CDPD. The reason of the slowness is merging norms without HDFS caching.

HDFS caching can not be enabled in the affected MRIT versions. Future MRIT releases will both allow controlling HDFS caching and will turn it on by default. For existing MRIT releases, the only known workaround is omitting norms. This disables length normalization for the field, saves some memory and improves MRIT execution times. Only full-text fields can benefit from norms. Norms are omitted for primitive (non-analyzed) types by default. (Norms were formerly also used for index-time boosting but this usage has been deprecated. Index-time boosting can be achieved using doc values fields instead.)

The downside of omitting norms is that document length will not play a role in result ranking. (With norms enabled, documents with a shorter matching field would be ranked higher than matching documents with a longer field.)

You can control norms in the schema using the omitNorms attribute in the fieldType elements. To eliminate the slowdown, you must add omitNorms="true" to all fieldType elements. It is also possible to selectively set this attribute on selected fields, which allows reducing the slowdown without completely eliminating it.

Indexing fails with socketTimeout

Starting from CDH 6.0, the HTTP client library used by Solr has a default socket timeout of 10 minutes. Because of this, if a single request sent from an indexer executor to Solr takes more than 10 minutes to be serviced, the indexing process fails with a timeout error.

This timeout has been raised to 24 hours. Nevertheless, there still may be use cases where even this extended timeout period proves insufficient.

If your MapreduceIndexerTool or HBaseMapreduceIndexerTool batch indexing jobs fail with a timeout error during the go-live (Live merge, MERGEINDEXES) phase (This means the merge takes longer than 24 hours).

Use the --go-live-timeout option where the timeout can be specified in milliseconds.

If the timeout occurs during Near real time (NRT) indexing, Cloudera suggests you try the following workarounds:

  • Check the batch size of your indexing job. Sending too large batches to Solr might increase the time needed on the Solr server to process the incoming batch.
  • If your indexing job uses deleteByQuery requests, consider using deleteById wherever possible as deleteByQuery involves a complex locking mechanism on the Solr side which makes processing the requests slower.
  • Check the number of executors for your Spark Crunch Indexer job. Too many executors can overload the Solr service. You can configure the number of executors by using the --mappers parameter
  • Check that your Solr installation is correctly sized to accommodate the indexing load, making sure that the number of Solr servers and the number of shards in your target collection are adequate.
  • The socket timeout for the connection can be configured in the morphline file. Add the solrClientSocketTimeout parameter to the solrLocator command


      collection : test_collection 
      zkHost : "zookeeper1.example.corp:2181/solr" 
    # 10 minutes in milliseconds 
      solrClientSocketTimeout: 600000  
      # Max number of documents to pass per RPC from morphline to Solr Server
      # batchSize : 10000
Splitshard operation on HDFS index checks local filesystem and fails

When performing a shard split on an index that is stored on HDFS, SplitShardCmd still evaluates free disk space on the local file system of the server where Solr is installed. This may cause the command to fail, perceiving that there is no adequate disk space to perform the shard split.

Run the following command to skip the check for sufficient disk space altogether:
  • On nonsecure clusters:
    curl 'http://$[***SOLR_SERVER_HOSTNAME***]:8983/solr/admin/collections?action=SPLITSHARD&collection=[***COLLECTION_NAME***]&shard=[***SHARD_TO_SPLIT***]&skipFreeSpaceCheck=true'
  • On secure clusters:
    curl -k -u : --negotiate 'http://$[***SOLR_SERVER_HOSTNAME***]:8985/solr/admin/collections?action=SPLITSHARD&collection=[***COLLECTION_NAME***]&shard=[***SHARD_TO_SPLIT***]&skipFreeSpaceCheck=true'

Replace [***SOLR_SERVER_HOSTNAME***] with a valid Solr server hostname, [***COLLECTION_NAME***] with the collection name, and [***SHARD_TO_SPLIT***] with the ID of the to split.

To verify that the command executed succesfully, check overseer logs for a similar entry:

2021-02-02 12:43:23.743 INFO  ( [c:example s:shard1  ] o.a.s.c.a.c.SplitShardCmd Skipping check for sufficient disk space
Lucene index handling limitation
The Lucene index can only be upgraded by one major version. Solr 8 will not open an index that was created with Solr 6 or earlier.
There is no workaround, you need to reindex collections.
Solr service with no added collections causes the upgrade process to fail
Upgrade fails while performing the bootstrap collections step of the script with the error message:
Failed to execute command Bootstrap Solr Collections on service Solr
if there are no collections present in Solr.
If there are no collections added to it, remove the Solr service from your cluster before you start the upgrade.
Collection Creation No Longer Supports Automatically Selecting A Configuration If Only One Exists

Before CDH 5.5.0, a collection could be created without specifying a configuration. If no -c value was specified, then:

  • If there was only one configuration, that configuration was chosen.

  • If the collection name matched a configuration name, that configuration was chosen.

Search now includes multiple built-in configurations. As a result, there is no longer a case in which only one configuration can be chosen by default.

Explicitly specify the collection configuration to use by passing -c <configName> to solrctl collection --create.
CrunchIndexerTool which includes Spark indexer requires specific input file format specifications
If the --input-file-format option is specified with CrunchIndexerTool, then its argument must be text, avro, or avroParquet, rather than a fully qualified class name.
The file does not validate ZooKeeper and the NameNode on some operating systems.
The file uses the timeout function to determine if ZooKeeper and the NameNode are available. To ensure this check can be complete as intended, the determines if the operating system on which the script is running supports timeout. If the script detects that the operating system does not support timeout, the script continues without checking if the NameNode and ZooKeeper are available. If your environment is configured properly or you are using an operating system that supports timeout, this issue does not apply.
This issue only occurs in some operating systems. If timeout is not available, the quickstart continues and final validation is always done by the MapReduce jobs and Solr commands that are run by the quickstart.
Field value class guessing and Automatic schema field addition are not supported with the MapReduceIndexerTool nor with the HBaseMapReduceIndexerTool.
The MapReduceIndexerTool and the HBaseMapReduceIndexerTool can be used with a Managed Schema created via NRT indexing of documents or via the Solr Schema API. However, neither tool supports adding fields automatically to the schema during ingest.
Define the schema before running the MapReduceIndexerTool or HBaseMapReduceIndexerTool. In non-schemaless mode, define in the schema using the schema.xml file. In schemaless mode, either define the schema using the Solr Schema API or index sample documents using NRT indexing before invoking the tools. In either case, Cloudera recommends that you verify that the schema is what you expect, using the List Fields API command.
The Browse and Spell Request Handlers are not enabled in schemaless mode
The Browse and Spell Request Handlers require certain fields to be present in the schema. Since those fields cannot be guaranteed to exist in a Schemaless setup, the Browse and Spell Request Handlers are not enabled by default.
If you require the Browse and Spell Request Handlers, add them to the solrconfig.xml configuration file. Generate a non-schemaless configuration to see the usual settings and modify the required fields to fit your schema.
Enabling blockcache writing may result in unusable indexes.
It is possible to create indexes with solr.hdfs.blockcache.write.enabled set to true. Such indexes may appear corrupt to readers, and reading these indexes may irrecoverably corrupt indexes. Blockcache writing is disabled by default.
Users with insufficient Solr permissions may receive a "Page Loading" message from the Solr Web Admin UI.
Users who are not authorized to use the Solr Admin UI are not given a page explaining that access is denied to them, instead receive a web page that never finishes loading.
Using MapReduceIndexerTool or HBaseMapReduceIndexerTool multiple times may produce duplicate entries in a collection.
Repeatedly running the MapReduceIndexerTool on the same set of input files can result in duplicate entries in the Solr collection. This occurs because the tool can only insert documents and cannot update or delete existing Solr documents. This issue does not apply to the HBaseMapReduceIndexerTool unless it is run with more than zero reducers.
To avoid this issue, use HBaseMapReduceIndexerTool with zero reducers. This must be done without Kerberos.
Deleting collections might fail if hosts are unavailable.
It is possible to delete a collection when hosts that host some of the collection are unavailable. After such a deletion, if the previously unavailable hosts are brought back online, the deleted collection may be restored.
Ensure all hosts are online before deleting collections.

Unsupported Features

The following Solr features are currently not supported in Cloudera Data Platform: