hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HBASE-1058) Prevent runaway compactions
Date Mon, 11 May 2009 23:10:45 GMT

    [ https://issues.apache.org/jira/browse/HBASE-1058?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12708255#action_12708255
] 

Andrew Purtell commented on HBASE-1058:
---------------------------------------

I don't like this whole approach because it can tie up all IPC handlers, but I'm not sure
how to do better. One option is to fail the insert with a TooManyStoreFilesException or something
like that and have the client do the waiting, with backoff, etc. However a badly behaved client
can still DoS the HRS. Or maybe it is actually fine to let the store files build up and just
be smarter during compaction about trying to do too much and blow out heap?

> Prevent runaway compactions
> ---------------------------
>
>                 Key: HBASE-1058
>                 URL: https://issues.apache.org/jira/browse/HBASE-1058
>             Project: Hadoop HBase
>          Issue Type: Bug
>            Reporter: stack
>            Assignee: Andrew Purtell
>            Priority: Blocker
>             Fix For: 0.19.2, 0.20.0
>
>         Attachments: hbase-1058-2-v1.patch, hbase-1058-v2.patch, hbase-1058-v4.patch,
hbase-1058.patch
>
>
> A rabid upload will easily outrun our compaction ability dropping flushes faster than
we can compact them up.  Fix.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message