guihecheng commented on pull request #2290: URL: https://github.com/apache/ozone/pull/2290#issuecomment-852690836
> Thanks @guihecheng for working on this. Two general comments, > > 1. The config "hdds.datanode.failed.volumes.tolerated" takes failed metadata volumes into consideration. I would suggest use seperate properties to control the metadata volumes and data volumes. I'm fine with this separation, for example we could have 'hdds.datanode.failed.data.volumes.tolerated' and 'hdds.datanode.failed.meta.volumes.tolerated' and check them each. > 2. Would it be more clear that using different MutableVolumeSet to hold the data volume set and meta volume set? Yes, I think they could be in separate VolumeSets, then I think we could extract the periodic disk checker into HddsVolumeChecker and let these 2 VolumeSets register there volumes for check, then we retain a single checker for different kinds of volumes. Then it should be easy to add the potential rocksdb volumes mentioned below, in the future. > Please also do consider if the current abstration is enough if rocksdbs are combined into one instance and saved on a different disk than current data disk in future. -- 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. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
