Skip to main content
Solved

Worker group version compatibility with leader during network issues

  • Nov 4, 2026 11:53 AM UTC
  • 5 replies
  • 2 views

This message originated from Cribl Community Slack.
Click here to view the original link.

I have a worker group that is disconnected due to a network issue. Before that gets resolved, how many versions behind the leader can that group be before it becomes a problem?

Best answer by Brandon McCombs

This depends because there are a couple levels of interaction.
  • the leader builds config bundles based off the files it has locally so if a bundle is created with a newer software version on the leader and that gets pushed to workers the workers simply won't know what to do with the configuration or it could possibly cause bigger issues like we saw recently with the Event Breaker function used in pipelines that had adverse effect on workers running an older version than what was on the leader.
  • the cluster communications. This is based on a specific protocol but that protocol rarely sees any changes made to it.
Officially we support different patch and minor versions on leader vs worker, issues can arise unexpectedly such as the issue I mentioned above with the event breaker.
View original

5 replies

Jon Rust
Forum|alt.badge.img
  • Employee
  • Nov 4, 2026 11:53 AM UTC

Jon Rust
Forum|alt.badge.img
  • Employee
  • Nov 4, 2026 11:53 AM UTC
> • Minor Versions: Cribl supports Worker Groups or distributed instances that are one minor version behind the Leader Node. For example, a Worker Group on 4.5.1 and a Leader is on 4.6.0. >

Thank you! I've been looking for this for a while before asking. Must either be to early in the morning or too late in the week. :blob-heart:

This depends because there are a couple levels of interaction.
  • the leader builds config bundles based off the files it has locally so if a bundle is created with a newer software version on the leader and that gets pushed to workers the workers simply won't know what to do with the configuration or it could possibly cause bigger issues like we saw recently with the Event Breaker function used in pipelines that had adverse effect on workers running an older version than what was on the leader.
  • the cluster communications. This is based on a specific protocol but that protocol rarely sees any changes made to it.
Officially we support different patch and minor versions on leader vs worker, issues can arise unexpectedly such as the issue I mentioned above with the event breaker.

Our deployment is currently on 4.18.2 and we're looking to upgrade to the latest. If we plan on upgrading the disconnected group as soon as the connection is restored, should we even worry about problem upgrading today?