search

puppetlabs / puppetserver-helm-chart

The Helm Chart for Puppet Server
Apache License 2.0
52 stars 56 forks source link

Fix typo in PUPPETDB_JAVA_ARGS #201

Closed qdii closed 11 months ago

qdii commented 11 months ago

The proper flag is -Xloggc (see https://docs.oracle.com/javacomponents/jrockit-hotspot/migration-guide/logging.htm)

CLAassistant commented 11 months ago

CLA assistant check
All committers have signed the CLA.

ldaneliukas commented 11 months ago

The proper flag is -Xloggc (see https://docs.oracle.com/javacomponents/jrockit-hotspot/migration-guide/logging.htm)

-Xloggc is not used intentionally, as it was deprecated since Java 9 and using it logs a deprecation warning in PuppetDB logs indicating that it ignores it and uses Xlog:gc instead.

qdii commented 11 months ago

Hey Linas,

I didn't know about this deprecation, thanks for pointing it out.

However without this fix, installing the helm chart led to the process crash-looping in my cluster. Adding this fix to my values.yaml made it start fine.

Is it possible that the java runtime in the image itself is too old?

ldaneliukas commented 11 months ago

Hey Linas,

I didn't know about this deprecation, thanks for pointing it out.

However without this fix, installing the helm chart led to the process crash-looping in my cluster. Adding this fix to my values.yaml made it start fine.

Is it possible that the java runtime in the image itself is too old?

Are you positive that the problem was due to this? As I'm constantly redeploying a solution via this chart in the past few days and everything starts properly. Do you have any error logs that it causes?

The use of this was added specifically due to PuppetDB container logging a deprecation warning and saying that it will use this even if you specify Xloggc - you should see this warning in your logs if you change it to Xloggc.

qdii commented 11 months ago

Just to be sure, I'll try to remove the remove the line from values.yaml tonight, re-deploy the chart and add screenshots.

However, I also noticed that I was using version 8.2.0 of the chart. While 8.2.1 is documented in CHANGELOG.md, I couldn't upgrade to 8.2.1 because it's not been pushed to Artifact hub. This means that a number of fixes are not present, and maybe that's the cause of the difference in behavior between your tests and mine.

I'd recommend also pushing 8.2.1 to Artifact hub, even though 9.0.0 was pushed today, because 8.2.0 is broken — two image names are missing the ghcr.io repository prefix —.

ldaneliukas commented 11 months ago

@qdii the same change has now been merged in the containers-puppetdb entrypoint is it should really cause now issues. If you want, we can remove it here, since it is now the default anyway.

qdii commented 11 months ago

Ah thanks! I totally forgot to check. But happy to hear it's solved :) I'll delete it.