gortiz opened a new pull request, #17649: URL: https://github.com/apache/pinot/pull/17649
In https://github.com/apache/pinot/pull/17576 we detected an issue that proves we cannot downgrade from 1.4 or 1.5 to 1.3 without downtime, as MSE queries may be broken due to an incompatibility in the stage stats. It is important to know the issue is not introduced by https://github.com/apache/pinot/pull/17576 but detected by it. The main issue is that in 1.4 we introduced a cluster listener that detects whether the cluster is homogeneous or not, and in case it is not, MSE doesn't send stage stats. We did that because in 1.3 we detected a bug and the only way to fix it was to change the stats in 1.4 in a way that 1.3 fails when they are received. This worked fine when upgrading the system, but the listener is based on a version Helix property we started to set in 1.4. When rolling back to 1.3 that property is not changed (as it is unknown by 1.3), so the other nodes see the old value, which means they think the cluster is in fact homogeneous. Fixing this issue may be complex and given we don't expect people to downgrade from 1.5 to 1.3, we think it is just better to remove that check here. As always, it is recommended to upgrade one version at the time, so in case you need to downgrade from 1.5, you should go to 1.4, which is still tested. -- 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]
