1. We are moving from 4.X to 6.6. 2. Changed the schema - adding the version etc nothing major. 3. Full re-index of documents into the cluster - so this is not a migration. 4. Changed the the JVM parameter from 12GB to 16GB and did a restart. 5. Replicates go into recovery which fails to complete after many hours. They still respond to queries but the /update POST from the replicates fails with the 500 server error and a stack trace because of the number format failure.
My other cluster does not reuse any nodes. The restart went as expected with the JVM change. Al ________________________________ From: Erick Erickson <erickerick...@gmail.com> Sent: Thursday, September 21, 2017 8:25:32 AM To: solr-user Subject: Re: Replicates not recovering after rolling restart Hmmm, I didn't ask what version you're upgrading _from_. 5 years ago would be Solr 4. Are you replacing Solr 5 or 4? I'm guessing 5, but want to check unlikely possibilities. Next question: I'm assuming all your nodes have been upgraded to Solr 6, right? Best, Erick On Wed, Sep 20, 2017 at 7:18 PM, Bill Oconnor <bocon...@plos.org> wrote: > I have no clue where that number comes from it does not seem to be in the > actual post to the leader as seen in my tcpdump. It is mystery. > > ________________________________ > From: Walter Underwood <wun...@wunderwood.org> > Sent: Wednesday, September 20, 2017 7:00:53 PM > To: solr-user@lucene.apache.org > Subject: Re: Replicates not recovering after rolling restart > > >> On Sep 20, 2017, at 6:15 PM, Bill Oconnor <bocon...@plos.org> wrote: >> >> I restart using the standard "sudo service solr start/stop" > > You might look into what that actually does. > > wunder > Walter Underwood > wun...@wunderwood.org > http://observer.wunderwood.org/ (my blog) [https://wunderwood.files.wordpress.com/2017/02/diva.png?w=32]<http://observer.wunderwood.org/> Most Casual Observer<http://observer.wunderwood.org/> observer.wunderwood.org >