Solr MetadataTracker cause OOM

[vincent-kali]

Hi,

We're running alf5.2g CE / solr4 with less than 100K nodes in repo, and 12GB of java heap size, 16GB of system RAM.

Since somes days, we noticed a huge memory consumption and OOM errors. We were not able to link this behaviour with any particular change on the system. After a memory dump/analyse, we have seen that SolrTrackingPool-alfresco-MetadataTracker threads and org.alfresco.repo.web.scripts.solr.NodesMetaDataGet$FreemarkerNodeMetaData webscripts are consuming 12GB of JVM !

Indexes in solr looks OK, but the MetadataTracker seems to become crazy......

Did somebody experienced similar problem ?

Thanks for your help,

Vincent

[afaust]

I guess this is the same question / topic as https://hub.alfresco.com/t5/alfresco-content-services-forum/alfresco-consuming-unreferenced-memory/m...

[vincent-kali]

Not really, we faced a very high memory/cpu consumption during hours (we finally understood that a FIX command on solr indexes have been run). Looks like it rebuild the whole indexes. Is the FIX command on solr supposed to overload the alf tracker ?

[angelborroy · Accepted solution]

The FIX command is performed by MetadataTracker, so it has for sure impact in the overload of alf tracker.

[vincent-kali]

Ok thanks. It seems that we're using groups in a way that alfresco doesn't support... I'll report this in a separate thread.