hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Varun Sharma (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-8389) HBASE-8354 DDoSes Namenode with lease recovery requests
Date Mon, 22 Apr 2013 06:35:16 GMT

    [ https://issues.apache.org/jira/browse/HBASE-8389?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13637792#comment-13637792
] 

Varun Sharma commented on HBASE-8389:
-------------------------------------

Hi Ted,

+1 on patch for 0.94

This basically retains the old behaviour prior to hbase 7878. For people who run clusters
with tight hdfs timeouts and know that lease/block recovery for them shall occur within x
seconds, they can make use of the configurable retry interval to wait for the real recovery
to happen.

I feel that HBase 7878 does not cause data loss for hbase but I am not sure. My theory is
that for HBase, size of 1 WAL is < size of one HDFS block before it is rolled. Hence each
rolled WAL has one block and one on which the lease is being held contains 1 block under_recovery/under_construction
- that file has a size=0 in the namenode until the lease recovery is complete - this is because
there are no finalized blocks for the WAL. So, when we do the following, try to replay the
WAL without lease recovery being complete, we get a file of size 0 from the namenode.

>From the region server logs, it seems that we do not take size 0 for the file as the truth
but instead treat it as a failure until we get a WAL file size > 0.

However, if the size of the WAL is > HDFS block size then it is possible that some HDFS
blocks have been finalized and we get a file size >= 0 because of the finalized blocks.
In that case we could end up throwing away the last block belonging to the WAL. Maybe that
is why this was observed for accumulo but not for HBase.
                
> HBASE-8354 DDoSes Namenode with lease recovery requests
> -------------------------------------------------------
>
>                 Key: HBASE-8389
>                 URL: https://issues.apache.org/jira/browse/HBASE-8389
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Varun Sharma
>            Assignee: Varun Sharma
>            Priority: Critical
>             Fix For: 0.94.8
>
>         Attachments: 8389-0.94.txt, 8389-trunk-v1.txt, 8389-trunk-v2.patch, nn1.log,
nn.log, sample.patch
>
>
> We ran hbase 0.94.3 patched with 8354 and observed too many outstanding lease recoveries
because of the short retry interval of 1 second between lease recoveries.
> The namenode gets into the following loop:
> 1) Receives lease recovery request and initiates recovery choosing a primary datanode
every second
> 2) A lease recovery is successful and the namenode tries to commit the block under recovery
as finalized - this takes < 10 seconds in our environment since we run with tight HDFS
socket timeouts.
> 3) At step 2), there is a more recent recovery enqueued because of the aggressive retries.
This causes the committed block to get preempted and we enter a vicious cycle
> So we do,  <initiate_recovery> --> <commit_block> --> <commit_preempted_by_another_recovery>
> This loop is paused after 300 seconds which is the "hbase.lease.recovery.timeout". Hence
the MTTR we are observing is 5 minutes which is terrible. Our ZK session timeout is 30 seconds
and HDFS stale node detection timeout is 20 seconds.
> Note that before the patch, we do not call recoverLease so aggressively - also it seems
that the HDFS namenode is pretty dumb in that it keeps initiating new recoveries for every
call. Before the patch, we call recoverLease, assume that the block was recovered, try to
get the file, it has zero length since its under recovery, we fail the task and retry until
we get a non zero length. So things just work.
> Fixes:
> 1) Expecting recovery to occur within 1 second is too aggressive. We need to have a more
generous timeout. The timeout needs to be configurable since typically, the recovery takes
as much time as the DFS timeouts. The primary datanode doing the recovery tries to reconcile
the blocks and hits the timeouts when it tries to contact the dead node. So the recovery is
as fast as the HDFS timeouts.
> 2) We have another issue I report in HDFS 4721. The Namenode chooses the stale datanode
to perform the recovery (since its still alive). Hence the first recovery request is bound
to fail. So if we want a tight MTTR, we either need something like HDFS 4721 or we need something
like this
>   recoverLease(...)
>   sleep(1000)
>   recoverLease(...)
>   sleep(configuredTimeout)
>   recoverLease(...)
>   sleep(configuredTimeout)
> Where configuredTimeout should be large enough to let the recovery happen but the first
timeout is short so that we get past the moot recovery in step #1.
>  

--
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

Mime
View raw message