Erick Erickson created SOLR-9799:
------------------------------------
Summary: Deleting a collection doesn't completely remove all
traces from the Solr nodes if RELOAD failed due to configuration errors
Key: SOLR-9799
URL: https://issues.apache.org/jira/browse/SOLR-9799
Project: Solr
Issue Type: Bug
Security Level: Public (Default Security Level. Issues are Public)
Reporter: Erick Erickson
Priority: Minor
Jerome Yang from the user's list asked about this, I verified in 6x. Steps to
reproduce:
- create a one-shard, leader-only collection on a single Solr node
- push a flawed configuration to Zookeeper (I forced a syntax error in
managed-schema)
- RELOAD the collection, see the error.
- DELETE the collection
- fix and push the config to ZK
- try to CREATE the same collection again error: Core with name
'eoe_shard1_replica1' already exists.
- restart the Solr instance
- you can successfully CREATE the collection now.
NOTE: If you push the corrected configs and don't DELETE, you can RELOAD the
collection without restarting Solr.
Curiously, the core is actually _not_ on disk, I'm guessing that it's in the
CoreDescirptor list still or some such.
Thanks for reporting Jerome!
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
