gortiz opened a new issue, #8944:
URL: https://github.com/apache/pinot/issues/8944
Default values provided by Pinot Chart are not the best. Issues I’ve found:
1. At least in the server, logs are sent to /opt/pinot/logs but cannot be
read with kubectl logs
2. CPU quotas or shares are not defined, as a result, Java 11 decides to
only use a single CPU by default
* This is bad per se, but it is specially bad when G1GC is used, as it
is explicitly specified by the chart.
* This JVM behaviour when no shares are specified seems to be changed
several times in the past. For example[[JDK-8279484]
Runtime.availableProcessors reports incorrect processor count - Java Bug
System](https://bugs.openjdk.org/browse/JDK-8279484) fixes it in Java 19
We should either define some cpu limits in the deployments/statefulsets or
at least provide -XX:ActiveProcessorCount to use at least 2 CPUs.
More info:
* [Finding Ideal JVM Thread Pool Size With Kubernetes and
Docker](https://mucahit.io/2020/01/27/finding-ideal-jvm-thread-pool-size-with-kubernetes-and-docker/)
* [[JDK-8281181] Do not use CPU Shares to compute active processor count -
Java Bug System](https://bugs.openjdk.org/browse/JDK-8281181)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
