[
https://issues.apache.org/jira/browse/BOOKKEEPER-345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13452885#comment-13452885
]
Vinay commented on BOOKKEEPER-345:
----------------------------------
Thanks Sijie, for review.
{quote}the case is: there is no enough space to store an entry, but the
ledgerDirsManager doesn't find that the disk is out-of-usage. (user configure a
larger enough threshold).{quote}
Yes you are right. It can corrupt the entry log. This I had considered in the
initial patch, with some extra synchronization and storing the duplicate
entries. Since after introducing the disk checker daemon, I thought this is not
required, because daemon will check disk every 10 seconds by default and it
will detect the disk usage above threshold well ahead before actually gets to
that stage of corruption. I think here we need to recommend user to use default
of 95% of threshold.
{quote}And I think it would be better to add a specific exception for writes to
readonly bookie in client-side.{quote}
Yes, these changes will be available as part of BOOKKEEPER-347
Other comments I will address and post a new patch soon.
> Detect IOExceptions on entrylogger and bookie should consider next ledger
> dir(if any)
> -------------------------------------------------------------------------------------
>
> Key: BOOKKEEPER-345
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-345
> Project: Bookkeeper
> Issue Type: Sub-task
> Components: bookkeeper-server
> Affects Versions: 4.1.0
> Reporter: Rakesh R
> Assignee: Vinay
> Fix For: 4.2.0
>
> Attachments: BOOKKEEPER-345.patch, BOOKKEEPER-345.patch,
> BOOKKEEPER-345.patch, BOOKKEEPER-345.patch, BOOKKEEPER-345.patch
>
>
> This jira to detect IOExceptions in "EntryLogger", and will iterate over all
> the configured ledger(s) on IOException. Finally if no writable dirs
> available, will move bookie to r-o mode(if this mode is enabled).
> By default(r-o mode will be disabled) the bookie will shutdown if no writable
> disk available.
--
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
