I remember tat I need to empty my queue before restart the cluster too. This bug is a little scary because if you have a schedule system to deletes collections on the fly your cluster can blow and you don't know why …
-- Yago Riveiro Sent with Sparrow (http://www.sparrowmailapp.com/?sig) On Thursday, February 27, 2014 at 11:58 AM, Oliver Schrenk wrote: > Interesting. I deleted all old collections, configs, and clusterstate.json > from Zookeeper and I still had that problem. > > I’m quite new to Zookeeper, so some of what I say might be wrong. It seems > there were some outstanding changes in Zookeeper. Or at least I found some > queue items in a queue node. I deleted all of them and was able to start my > cluster. > > > On 27 Feb 2014, at 12:48, Yago Riveiro <yago.rive...@gmail.com > (mailto:yago.rive...@gmail.com)> wrote: > > > I had some problems with DELETE action too. > > > > I reported this some time ago > > https://issues.apache.org/jira/browse/SOLR-5559 > > > > The overseer fail to delete a collection and the solr cluster becomes > > unstable, I reloaded my boxes and my cluster never went online. > > > > After some debug, I found some shard folder with the name of the collection > > that I was delete previously. I deleted it and the cluster went online > > again. > > > > You will need to remove all entries of deleted collection manually on > > clusterstate.json (if exists any). > > > > > > -- > > Yago Riveiro > > Sent with Sparrow (http://www.sparrowmailapp.com/?sig) > > > > > > On Thursday, February 27, 2014 at 10:42 AM, Oliver Schrenk wrote: > > > > > Hi, > > > > > > Upgraded a small cluster from 4.3.1 to 4.7 in Solr Cloud mode. > > > > > > I deleted the old data, replaced the solr.xml with the example solr.xml > > > with auto-discovery, but it seems there is still some old data somewhere, > > > probably in Zookeeper that keep my machine from starting. > > > > > > It loops over the same log message over and over > > > > > > 2014-02-27 11:00:49,011 INFO o.a.s.c.c.ZkStateReader [Thread-15] Updating > > > cloud state from ZooKeeper... 2014-02-27 11:00:49,012 INFO > > > o.a.s.c.Overseer$ClusterStateUpdater [Thread-15] Removing collection: > > > collection1 shard: 2 from clusterstate 2014-02-27 11:00:49,012 ERROR > > > o.a.s.c.Overseer$ClusterStateUpdater [Thread-15] Exception in Overseer > > > main queue loop org.apache.solr.common.SolrException: Could not find > > > collection:collection1 at > > > org.apache.solr.common.cloud.ClusterState.getCollection(ClusterState.java:175) > > > at > > > org.apache.solr.cloud.Overseer$ClusterStateUpdater.removeShard(Overseer.java:801) > > > at > > > org.apache.solr.cloud.Overseer$ClusterStateUpdater.processMessage(Overseer.java:230) > > > at > > > org.apache.solr.cloud.Overseer$ClusterStateUpdater.run(Overseer.java:187) > > > at java.lang.Thread.run(Thread.java:722) > > > > > > How can I delete that false information? > > > > > > Regards > > > Oliver > > > > > > > > > >
