hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enis Soztutar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-7709) Infinite loop possible in Master/Master replication
Date Sat, 09 Mar 2013 00:03:14 GMT

    [ https://issues.apache.org/jira/browse/HBASE-7709?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13597711#comment-13597711
] 

Enis Soztutar commented on HBASE-7709:
--------------------------------------

I like option #2 better than this. It is more simpler. Jeff's idea is good, but has the problem
of dealing with the topology changes. If the topology changes in a way to make the normal
route to a cluster longer, than all the updates afterwards will be dropped unless we somehow
clear the cached mappings. This brings in an operational burden of cleaning the caches of
downstream clusters, once the admin changes the topology upstream. 
{code}
A -> B <-> C is changed to A -> B -> D -> C -> B 
{code}

Orthogonal to this, we also should be dropping the edits at the replication source, not the
sink. We are doubling the network cost in cyclic cases. #2 also helps with this condition,
because we can detect the sink cluster's id, and filter out. 

We can do a similar dynamic dictionary encoding for storing set of cluster ids. We can do
it as a follow up optimization.


                
> Infinite loop possible in Master/Master replication
> ---------------------------------------------------
>
>                 Key: HBASE-7709
>                 URL: https://issues.apache.org/jira/browse/HBASE-7709
>             Project: HBase
>          Issue Type: Bug
>          Components: Replication
>    Affects Versions: 0.95.0, 0.94.6
>            Reporter: Lars Hofhansl
>            Assignee: Jeffrey Zhong
>             Fix For: 0.95.0, 0.94.7
>
>
> We just discovered the following scenario:
> # Cluster A and B are setup in master/master replication
> # By accident we had Cluster C replicate to Cluster A.
> Now all edit originating from C will be bouncing between A and B. Forever!
> The reason is that when the edit come in from C the cluster ID is already set and won't
be reset.
> We have a couple of options here:
> # Optionally only support master/master (not cycles of more than two clusters). In that
case we can always reset the cluster ID in the ReplicationSource. That means that now cycles
> 2 will have the data cycle forever. This is the only option that requires no changes
in the HLog format.
> # Instead of a single cluster id per edit maintain a (unordered) set of cluster id that
have seen this edit. Then in ReplicationSource we drop any edit that the sink has seen already.
The is the cleanest approach, but it might need a lot of data stored per edit if there are
many clusters involved.
> # Maintain a configurable counter of the maximum cycle side we want to support. Could
default to 10 (even maybe even just). Store a hop-count in the WAL and the ReplicationSource
increases that hop-count on each hop. If we're over the max, just drop the edit.

--
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