[
https://issues.apache.org/jira/browse/SOLR-5559?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yago Riveiro updated SOLR-5559:
-------------------------------
Description:
I think that I found a bug in DELETE collectionAPI command.
Environment:
- N boxes, the number is not important.
- A collection with N shard spreed over the N boxes.
- Solr.xml old style.
I ran the command as
http://localhost:8983/sorl/admin/collections?action=DELETE&name=CollectionX
The command return a 200 all was cleaned and in theory the collection was
removed ... but for some reason, one of the boxes doesn't delete the references
of CollectionX from the solr.xml and the folders of cores still exists. The
clusterstate.json doesn't have the CollectionX and the /collections doesn't
show the collectionX either.
This result of this situation is an exception in overseer queue loop like this:
{{org.apache.solr.cloud.Overseer$ClusterStateUpdater; Exception in Overseer
main queue loop}}
This exception stuck the queue and stoping the cluster. I think that is easy
replicate it with a test case.
I think that before to send an ok in DELETE command we must ensure that nothing
about this collection still existing on the cluster.
was:
I think that I found a bug in DELETE collectionAPI command.
Environment:
- N boxes, the number is not important.
- A collection with N shard spreed over the N boxes.
- Solr.xml old style.
I ran the command as
http://localhost:8983/sorl/admin/collections?action=DELETE&name=CollectionX
The command return a 200 all was cleaned and in theory the collection was
removed ... but for some reason, one of the boxes doesn't delete the references
of CollectionX from the solr.xml and the folders of cores still exists. The
clusterstate.json doesn't have the CollectionX and the /collections doesn't
show the collectionX either.
This result of this situation is an exception in overseer queue loop like this:
{{org.apache.solr.cloud.Overseer$ClusterStateUpdater; Exception in Overseer
main queue loop}}
This exception stuck the queue and stoping the cluster. I think that is easy
replicate it with a test.
I think that before to send an ok in DELETE command we must ensure that nothing
about this collection still existing on the cluster.
> DELETE collection command doesn't works in some cases
> -----------------------------------------------------
>
> Key: SOLR-5559
> URL: https://issues.apache.org/jira/browse/SOLR-5559
> Project: Solr
> Issue Type: Bug
> Affects Versions: 4.6
> Reporter: Yago Riveiro
>
> I think that I found a bug in DELETE collectionAPI command.
> Environment:
> - N boxes, the number is not important.
> - A collection with N shard spreed over the N boxes.
> - Solr.xml old style.
>
> I ran the command as
> http://localhost:8983/sorl/admin/collections?action=DELETE&name=CollectionX
> The command return a 200 all was cleaned and in theory the collection was
> removed ... but for some reason, one of the boxes doesn't delete the
> references of CollectionX from the solr.xml and the folders of cores still
> exists. The clusterstate.json doesn't have the CollectionX and the
> /collections doesn't show the collectionX either.
> This result of this situation is an exception in overseer queue loop like
> this:
> {{org.apache.solr.cloud.Overseer$ClusterStateUpdater; Exception in Overseer
> main queue loop}}
> This exception stuck the queue and stoping the cluster. I think that is easy
> replicate it with a test case.
> I think that before to send an ok in DELETE command we must ensure that
> nothing about this collection still existing on the cluster.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
