[
https://issues.apache.org/jira/browse/BOOKKEEPER-293?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13527724#comment-13527724
]
Uma Maheswara Rao G commented on BOOKKEEPER-293:
------------------------------------------------
Hi Ivan, Thanks a lot for the patch. Recent days I did not get proper time to
think on it, due to our deadlines/ other works. :-(
Some thoughts I am having on it, But your approach is straight forward and ok
for now to include in 4.2 release.
Here the problem with current approch may be is, It will do remote read of all
ledgers once in a day. This will consume the NW bandwidth and disturb other
processes or BK perf at that time right.
To avoid this remote read, how about implementing something like, Auditor will
be just co-ordinator for this work. When the period triggers, Auditor will just
send a requests to all RWs for scanning their local BK entries. Once the
scannig done, RW will send back the results to Auditor. After that Auditor will
publish based on the results from RW. RW anyway can pick the ledgers
underreplicated and do replication. Scanning before replication toout the
actual entry missings will happen only if there were real failures seen right.
So, this scanning may be acceptable. But I am worried bit about the full
ledgers scaning for disk failures. This can be a future work if you think this
is correct to do.
Other than that, patch looks good at high level. I will do further reviews in
detail and post the comments in some time today.
Thanks,
Uma
> Periodic checking of ledger replication status
> ----------------------------------------------
>
> Key: BOOKKEEPER-293
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-293
> Project: Bookkeeper
> Issue Type: Sub-task
> Components: bookkeeper-auto-recovery
> Reporter: Ivan Kelly
> Assignee: Ivan Kelly
> Fix For: 4.2.0
>
> Attachments:
> 0002-BOOKKEEPER-293-Periodic-checking-of-ledger-replicati.patch
>
>
> We should periodically check all ledgers to make sure that they are fully
> replicated.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
