[
https://issues.apache.org/jira/browse/CASSANDRA-20565?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17945442#comment-17945442
]
C. Scott Andreas commented on CASSANDRA-20565:
----------------------------------------------
For operational issues or questions regarding the database, please reach out
via the user mailing list or #cassandra channel in ASF Slack. Details on how to
join these are here: [https://cassandra.apache.org/_/community.html]
The database's bug tracker isn't the proper location for this report, though.
If your OOM happens to be caused by allocations of native memory during the
repair and you're setting use_offheap_merkle_trees=true, you may try setting
this to false.
Please do feel free to continue discussion via one of the two avenues above,
though.
> Out of memory: Killed process 2925120 (java) total-vm: 98267864kB, anon-rss:
> 29601144kB, file-rss: 17412kB, shmem-rss: 0kB, UID: 989, pgtables: 114272kB,
> oom_score_adj: 0
> --------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-20565
> URL: https://issues.apache.org/jira/browse/CASSANDRA-20565
> Project: Apache Cassandra
> Issue Type: Bug
> Components: Tool/nodetool
> Reporter: akshaygu
> Priority: Normal
> Attachments: Cassandra.yaml.txt, Cassandra_process.txt,
> nodetool_repair_log.txt
>
>
> Hi,
>
> We have Cassandra cluster with 9 nodes with Standard_D8s_v5/8vCPU-32GiB
> configuration. We have ordering orchestration application which uses
> Cassandra DB to store the services. We have migrated 40Mil services on
> Cassandra DB.
> Currently, we are running nodetool repair --full on Cassandra VMs. The job
> usually takes 7 to 9 hours but since 2 weeks the job is failing. The
> cassandra services getting crashed because OOM while Nodetool repair is
> running. It caused disruption in repair service.
> Initially we have below configuration then later we change it.
> *Old*
> -Xms16G
> -Xmx16G
> *New*
> -Xms8G
> -Xmx8G
> -XX:MaxDirectMemorySize=12g
>
> Still the Cassandra nodetool is failing. I have attached the logs from the
> day of failure and Cassandra configuration files.
> Kindly assist here.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
