High CPU usage on Platform appliance from Java and yarn - VCF Operations for Networks
Article ID: 410700
Updated On:
Products
Issue/Introduction
On a VCF Operations for Networks platform node, there is high CPU usage reported and possibly out of memory errors.
Utilizing SSH to log in to the platform node as support user and running the top command, shows results with significantly and consistently over 100% CPU usage for some users of the java service, like yarn, root and/or ubuntu.
From the web console, you may see out of memory errors:
Memory cgroup out of memory: Killed process ###### (java) total-vm:######kB, anon-rss:######kB, file-rss:######kB,shmem-rss:0kB, UID:### pgtables:####kB oom_score_ad j: 1000
Environment
VCF Operations for Networks 6.13
VCF Operations for Networks 6.14.x
VCF Operations for Networks 9.x
Resolution
To workaround this issue, reboot the platform node.
- If you have a clustered environment, use the procedure to shutdown your nodes documented in KB 314428 Best practices to shutdown VCF Operations for Networks Clustered deployments.
Post reboot of nodes if the issue still exists then this need to be further triage and debugged. Open a support case with Broadcom Support to review your VCF Operations for Networks deployment and refer to this KB. For more information, see Creating and managing Broadcom support cases.
Feedback
