cancel
Showing results for 
Search instead for 
Did you mean: 

Alfresco 4.2 - long long (almost neverending) loop "checking aux doc exists in index before we update it"

ktrls
Champ on-the-rise
Champ on-the-rise

Hi, have you ever seen this kind of log in Solr?

[solr.tracker.CoreTracker] [SolrTrackingPool-alfresco1] .. checking aux doc exists in index before we update it

I notice that Solr don't update its indexes till  it is not ended.  This is owful because all new data are not searchable by lucene search.

Why is there obsessive behavior?

Is there a "magic tip" to stop this activity? What is useful for? Is dangerous to stop it? Is it really mandatory?

I searched on google but I found only this: https://svn.alfresco.com/repos/alfresco-open-mirror/alfresco/HEAD/root/projects/solr/source/java/org...

and I don't understand what it means.

thanks in advance

4 REPLIES 4

josh_barrett
Confirmed Champ
Confirmed Champ

We have the same issue.  It is killing our index tracking.  We are using SOLR 1.4.  Going to be upgrading to 5.1.1 with SOLR4.   Hoping this will fix the issue.

Our model is very simple.   I am not sure what is causing this.   It got really bad when we got to around 10 million documents in our index.  

‌,  we reindexed and the issue seems to have stopped.  Or tracking isn't getting hung up anymore.   The indexes are also about 2/3 the size they were before (20G).   Hope this helps.

Thanks @Josh_Barrett _, we can't reindex shortly all because it means to stop access to the application for almost a week (we have indexes with hundreds of GB).

We have separated solr and alfresco so now they works on different tomcats and an Alfresco expert have made some optimization.

The situation is critical but now works.

Thanks for your feedback

We had the same problem with the reindex.  It took a week...

We stood up a new VM with a clean install of SOLR and pointed it to Alfresco.  

It reindexed on the side.  Once it was caught up we used our next open maintenance window to stop all of the SOLRs and copy the fresh indexes over to the other SOLR servers.   (took around 2 hours)

Once we started up the SOLRs we were good.  Haven't seen the issue since.