Github user sijie commented on the issue: https://github.com/apache/bookkeeper/pull/189 @jvrao @reddycharan one more question, most of the comments are around the performance on journal directories. but it turns out this change has nothing to do with the journal directories. The change applies only on the ledger directories and index directories. if that's the case, I don't see a reason to have this configuration. because random I/O anyway exists on ledger and index directories, there isn't any performance difference between configuring one directory per disk partition or multiple directories per disk partition. Filesystem I/O scheduling will handle either case in the same way. so this doesn't sound like an operation error as what @jvrao said. so back to the question, what is the problem that this pull request want to address? - if the problem is about the mis-reported disk metrics, this can be done smarter by de-duplicating the directories within ledger disks and report the per disk partition metrics. - if the problem is about performance, multiple ledger/index disks on same partition doesn't have any performance difference with one ledger/index disk per disk partition. this doesn't sound like an operation error that needs to be fixed. - if the problem is about performance on journal disks, I am fine with such setting. but this pull request need to improve to address the right problem.
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---