On 10/13/2016 2:50 AM, Angelo Gaverini wrote: > We use solr to index file inside an alfresco application, that is not > working good anymore and we don't understand why. > When the solr stop index in the logs we read this row: > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > > We tried to restart tomcat a lot of times and we observe this strange > behavior: > - solr start index, everytingh is running good and we can see 24 thread > working, log say: > > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco20] .. checking > for path change > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco9] .. checking > for path change > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco2] .. checking > for path change > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco12] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco3] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco1] .. updating > > - after some minutes one thread start writing "checking aux doc exists in > index before we update it" while other thread is working good > > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco3] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco1] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco15] .. checking > for path change > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco21] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco24] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco14] .. updating > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco9] .. updating > > - after some minutes no-one thread is working and in the log a can find > only the row: > > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it > INFO [solr.tracker.CoreTracker] [SolrTrackingPool-alfresco4] .. checking > aux doc exists in index before we update it
These logs did not come from Solr. If you need help with what this means, you're going to need to talk to whoever created the software, which is presumably whoever created Alfresco. We have no idea what this software is doing, or what those log messages mean. If you have proof that the priblem is in *Solr*, then show us evidence from Solr. Otherwise, you should be talking to somebody who knows about the software that created the log above. If they investigate the problem and believe that it is Solr, then they may be able to help you find the evidence you need to ask about on this list. > After this no-one index was created. > > I found then in the index directory the file > "lucene-85839584395834953-write.lock" and the creation date is the same as > the first row with the message "checking aux doc exists in index before we > update it" in the logs. The write lock is created by Lucene, by the part of the code that changes the index, to ensure that only one process (in this case, Solr, which is a Lucene application) can update the on-disk index. It sounds like Solr opens the index writer object once you request a change to the index, and probably keeps it open indefinitely after that unless you restart Solr. A core reload might also remove the lock until the next change, but without testing, I can't say for sure. That is an EXTREMELY low-level detail that you really don't need to be worried about. Solr and Lucene will handle it, and I am not aware of any bugs in that part of the system. It looks like somebody else is asking the list separately about this same problem, probably from your organization. They asked about deleting the write lock. Don't do that. I don't know what would happen, but it probably wouldn't be good. Thanks, Shawn
