[Deploying Sakai] [Building Sakai] Unhelpful error messages from search

Yuji Shinozaki ys2n at virginia.edu
Thu May 21 07:19:13 PDT 2009


Hey Dave,

We have found that local search indexing is fairly fragile when you  
are bringing nodes up and down.  Ian Boston and company had a method  
for fixing up the local indexes when a node restarts (by copying a  
"last known good" copy from another node).  We haven't gotten around  
to implementing their workaround, but I'll scare up the notes about it  
and forward them to you.

yuji
----


On May 21, 2009, at 8:51 AM, David Horwitz wrote:

> Hi David,
>
> What appears to have happened is that your node "ketchup" was shut  
> down
> while it was optimizing the shared index. It is likely that if you
> update the statuses in the db to
> "committed" the merging will continue. I would check though that all  
> the
> nodes mentions in the search_journal table are present in your shared
> index though
>
> David
>
>
> Adams, David wrote:
>> We recently did a reconfiguration of our Sakai deployment and when we
>> brought things back up, we noticed a lot of messages in the log from
>> org 
>> .sakaiproject.search.optimize.shared.impl.DbJournalOptimizationManage
>> r. Every ten seconds, it says:
>>
>> ERROR: This node already merging shared segments, index writer
>> ketchup:1242477119395   This node is currently optimizing the shared
>> segments,      This is an error as only one copy of this node  
>> should be
>> Active in the cluster
>>
>> Looking at the error in the source  
>> (DbJournalOptimizationManager.java in
>> search-impl) isn't any more helpful, but I did look at the table  
>> that is
>> triggering this message, SEARCH_JOURNAL, and we have 51 entries with
>> status 'merging-prepare' with indexwriter "ketchup:1242477119395" and
>> 795 entries with status 'committed' and indexwriter "ketchup".
>>
>> I haven't looked any deeper yet, but I am suspecting this is just a
>> matter of the database getting out of sync with reality for whatever
>> reason. Does anyone have any tips on safe ways to try to address
>> whatever the issue is? Or pointers to useful resources on  
>> understanding
>> how search works? I'm not sure what information in the wiki is  
>> reliable.
>> We're on the 2.5.4 tag if it makes a difference.
>>
>> David Adams
>> Director of Learning Systems Integration and Support
>> Virginia Tech Learning Technologies
>>
>> _______________________________________________
>> sakai-dev mailing list
>> sakai-dev at collab.sakaiproject.org
>> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
>>
>> TO UNSUBSCRIBE: send email to sakai-dev-unsubscribe at collab.sakaiproject.org 
>>  with a subject of "unsubscribe"
>>
> _______________________________________________
> production mailing list
> production at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/production
>
> TO UNSUBSCRIBE: send email to production-unsubscribe at collab.sakaiproject.org 
>  with a subject of "unsubscribe"

-----
Yuji Shinozaki
Sr. Technical Lead/Project Manager
University of Virginia
Advanced Technologies Group
ys2n at virginia.edu




More information about the production mailing list