You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by Rohit Harchandani <rh...@gmail.com> on 2013/03/21 17:08:27 UTC

Problem with Solr replication in solr 4.2

Hey,
Currently we are using solr 4.0 with a master slave setup. The data gets
indexed on the master and then we issue a fetchindex command to replicate
it on the slave. The slave has a postCommit listener which gets kicked off
when replication finishes and we depend on this listener to know whn
replication is done. But when I tried to do the same with 4.2, the commit
does not seem to be happening. Is this a known issue? is there any other
way to know that replication is done?
Also, initially when i tried solr 4.2, i noticed with this setup, i noticed
that with the fetchIndex command, the fields were downloaded to the temp
folder, but it was never pulled into the index directory on the slave. The
only file which made it was the lock file. This problem does not happen
anymore?
Thanks,
Rohit

Re: Problem with Solr replication in solr 4.2

Posted by Mark Miller <ma...@gmail.com>.
On Mar 21, 2013, at 12:08 PM, Rohit Harchandani <rh...@gmail.com> wrote:

> Hey,
> Currently we are using solr 4.0 with a master slave setup. The data gets
> indexed on the master and then we issue a fetchindex command to replicate
> it on the slave. The slave has a postCommit listener which gets kicked off
> when replication finishes and we depend on this listener to know whn
> replication is done. But when I tried to do the same with 4.2, the commit
> does not seem to be happening. Is this a known issue? is there any other
> way to know that replication is done?

Yeah, we stopped doing this commit because it changes some index meta data and shouldn't be necessary. I'd open a JIRA issue about being able to listen for replication finishing.

> Also, initially when i tried solr 4.2, i noticed with this setup, i noticed
> that with the fetchIndex command, the fields were downloaded to the temp
> folder, but it was never pulled into the index directory on the slave. The
> only file which made it was the lock file. This problem does not happen
> anymore?

I don't know, does it? Can you file a JIRA with instructions to replicate?

4.2.1 is about to go out, if we are quick, perhaps we can address something here if there is a problem.

- Mark

> Thanks,
> Rohit