gharris1727 commented on PR #14055: URL: https://github.com/apache/kafka/pull/14055#issuecomment-1646086669
> The question of plugins that are installed directly onto the classpath might be a bit tricky... let me know if I've missed something. My current understanding is that these plugins will continue to be available on the worker regardless of which discovery mode is used (with the possible exception of HYBRID_FAIL, if they prevent workers from starting up), but that they may not be visible in portions of the REST API (I'm thinking GET /connector-plugins, though there may be others). This is incorrect, and I think i've contributed to the confusion, sorry. To be very explicit: 1. The `plugin.discovery` configuration treats classpath and plugin.path the same. If a non-migrated plugin is on the classpath, then it will appear in HYBRID_WARN warnings, HYBRID_FAIL exceptions, and won't be usable with SERVICE_LOAD. It will not appear in GET /connector-plugins and attempting to use it will cause ClassNotFound or similar exceptions to be thrown. 2. The connect-plugin-path script will only migrate plugins included in it's arguments, leaving classpath completely unchanged. The `list` command will not show classpath plugins. This means someone may have HYBRID_FAIL enabled, and the connect-plugin-path script will not solve it, such as in unit tests, when everything is on the classpath. 4. If someone adds non-migrated plugins to the classpath (such as in a production deployment), they will need to pay additional attention to those plugins. They can either move them to `plugin.path`, upgrade them, or use `--plugin-location` to specify each one manually. -- 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: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
