swamirishi commented on PR #9313: URL: https://github.com/apache/ozone/pull/9313#issuecomment-3547785822
> Having local YAML for every snapshot have below concern, > > 1. for any update of yaml, need ensure, update done in new file, existing to be moved to backup and rename new one, delete backup. > 2. Any failure in above need a recovery during startup > 3. Reading 64k yaml file is slower for doing verification. > 4. A scheduler to verify the consistency of file or during startup > > My suggestion, > > * Can have another common local-rocksdb or snapshot-db for this, which handle all above implicitly. So this do not need scan of each file and its transaction safe implicitly present. > > related to redability, > > * value can be yaml content OR ozone ldb can support same. I don't think we need to be worried about the scale aspect here. Reading one yaml file would be order of 100 microseconds, so even if we have to read 64000 snapshots we are talking about few seconds at best. We can create a backburner jira for this and move this into a rocksdb later on. Once we are done with the defrag implementation. -- 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]
