[
https://issues.apache.org/jira/browse/BEAM-10281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17145321#comment-17145321
]
Tyson Hamilton commented on BEAM-10281:
---------------------------------------
workers 1-4 increased to 4 executors each. Worker #2 is running 4 jobs and has
plenty of memory still. Jobs:
{code:java}
beam_PreCommit_Java_Commit, beam_PreCommit_Java_Commit,
beam_PreCommit_Java_Commit, beam_PreCommit_Java_Commit,
beam_PreCommit_Python_Commit
{code}
Memory:
{code:java}
...@apache-ci-beam-jenkins-2:~$ free -g
total used free shared buff/cache available
Mem: 102 48 22 0 31 52
Swap: 0 0 0
{code}
> Jenkins Load Sharing Improvements
> ---------------------------------
>
> Key: BEAM-10281
> URL: https://issues.apache.org/jira/browse/BEAM-10281
> Project: Beam
> Issue Type: Bug
> Components: build-system
> Reporter: Tyson Hamilton
> Assignee: Tyson Hamilton
> Priority: P2
>
> Jenkins has 16 agents that run two executors on n1-highmem-16 (16 vCPUs, 104
> GB memory) vms on GCE. The nodes appear to be far underutilized (CPU, MEM,
> IO):
>
> With 12 jobs in the queue, a sampling of the agent VMs show:
> CPU hovers around 10-30% utilization
> Memory hovers around 2-15%
>
> These configurations should be revisited to either increase the number of
> executors per agent or decrease the VM size and increase the number of VMs.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
