If you have legacyCloud set as true, then when the various Solr instances come back up they re-construct their state. The "smoking gun" here is if the data is in clusterstate.json in your new Zookeeper.
If that's not operative, is it possible that you're still starting the embedded Zookeeper and getting data from _there_? Best, Erick On Thu, Mar 9, 2017 at 6:23 AM, Satya Marivada <satya.chaita...@gmail.com> wrote: > Hi, > > We had solr running on embedded zookeeper. Moved to external zookeeper, as > part of this setup on the same vm, had done a fresh solr distribution > setup, zookeeper distribution and created new solrdata folder to hold the > nodes. All the old folders are archived (zipped and backed up). What > wonders is the new deploy pointing to external zookeeper still shows the > old collections that were created on embedded zookeeper on the solr admin > console. How is that possible, when I did a clean fresh install of the solr > distribution and zookeeper distribution. Is that the old collection > information stored in another location as well apart from solr-6.3.0 > distribution package, zookeeper-3.4.9 package and the solrdata folder? > Would solr write into any other directories? > > Thanks, > Satya
