sodonnel commented on PR #11602: URL: https://github.com/apache/ozone/pull/11602#issuecomment-6264142347
The way this currently works is fundamental to the design on QC handling. In order to close the container, we must see all 3 origins and then the "best" is picked, with best meaning the one with most data. The others are then rightly removed. I don't think another solution that keeps extra copies around forever is the answer. Probably a reconciliation step which compares the contents of the containers and repairs the ones with lagging BCSID, as the difference should be very small. The reconciler is already there and can do the repair, but RM does not schedule any scans etc. Once we start keeping around extra copies (and we do unfortunately do this for unhealthy and QC Stuck containers) how do we account for quota, space used etc as all the formulas assume RATIS-3 == 3x the real space. If we starting having a lot of retained extra containers that will drift and be very hard to reconcile. There needs to be a "way out" for the extra copies - not just keep them forever. -- 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]
