[
https://issues.apache.org/jira/browse/SOLR-10108?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Erick Erickson updated SOLR-10108:
----------------------------------
Attachment: SOLR-10108.patch
Well, it works for me. If someone with a Windows setup could try the following
I'd be grateful:
Set up a SolrCloud instance, just enough to get it running and see the cloud
graph.
> 'bin/solr zk cp -r zk:/ file:/some/dir -z localhost:2181'
> Shut down Solr
> Re-initialize ZK (wipe all the data)
> 'bin/solr zk cp -r file:/some/dir/ zk:/ -z localhost:2181'
> start Solr again
If I'm lucky, that'll start Solr up seamlessly. The idea is that the above
should save the entire ZK state to a local drive and then restore it. We had a
case where the Solr nodes were still up but all the ZK data had been lost and
this would have been simpler than what we had to do.
This is something of a side-effect of making the recursive copy just work
to/from the root of ZK. There are problems with recursive copy in the first
place when it's not from some node below the root without this patch.
This patch also handles the fact that non-leaf nodes in Solr can have data and
properly saves/restores that data.
> bin/solr script recursive copy broken
> -------------------------------------
>
> Key: SOLR-10108
> URL: https://issues.apache.org/jira/browse/SOLR-10108
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Erick Erickson
> Assignee: Erick Erickson
> Attachments: SOLR-10108.patch
>
>
> cp /r zk:/ fails with "cannot create //whatever".
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
