lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shawn Heisey <apa...@elyograg.org>
Subject Re: Bringing Old Collections Up Again
Date Tue, 03 May 2016 15:04:15 GMT
On 5/2/2016 5:16 AM, Salman Ansari wrote:
> After several trials, it did start Solr on both machines but *non of the
> previous collections came back normally.* When I look at the admin page, it
> shows errors as follows
>
> *[Collection_name]_shard2_replica2:*
> org.apache.solr.common.SolrException:org.apache.solr.common.SolrException:
> Index locked for write for core '[Collection_name]_shard2_replica2'. Solr
> now longer supports forceful unlocking via 'unlockOnStartup'. Please verify
> locks manually!

I thought I had already sent this, but looks like it's still hanging
around.  Erick said much the same thing in his reply.

It sounds like Solr is being forcibly killed, not gracefully shut down,
when the machine is rebooted after installing updates.  When Solr is
forcibly killed, you probably need to find and delete a file named
"write.lock" in each of your index directories.  If that file exists
before Solr starts, Solr will not be able to lock the index and will not
load that index.  Restarting Solr after fixing the problem might be the
only solution.

You should not allow these systems to automatically reboot for updates,
or you should see if you can extend the amount of time that the reboot
will wait for processes to shut down normally before forcibly killing
them.  A graceful shutdown of a Solr process handling a lot of large
indexes can take quite a while.

A question for pondering:  Should the start script check for
index/write.lock files and fail to start (with a clear error message) if
any are found?  If we do that, we probably also need a "cleanlocks"
command on the solr script (and probably the init script) to delete
those files.

Thanks,
Shawn


Mime
View raw message