[
https://issues.apache.org/jira/browse/SOLR-11720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16277714#comment-16277714
]
Vitaly Lavrov commented on SOLR-11720:
--------------------------------------
Backup restore command is sent as
http://solr.host.com:8983/solr/admin/collections?action=RESTORE&name=offers&location=/backup&collection=offers&async=101&wt=json
The target collection 'offers' gets deleted right before restore command is
sent (so does configSet).
1. No commits are sent, only searches
2. Old searcher does not exist because collection is being restored
3.
Timeline after restore command has shown 'completed' status:
10 seconds after 448435467 documents
20 seconds after 454947732 documents
30 seconds after 460556612 documents
40 seconds after 466567718 documents
50 seconds after 472862588 documents
60 seconds after 472862588 documents
70 seconds after 479985875 documents (final number, the correct one)
Number of documents are measured with
http://solr.host.com:8983/api/c/offers/query?q=*:*
4. Searches go directly to solr cloud
> Backup restore command completes before index is ready
> ------------------------------------------------------
>
> Key: SOLR-11720
> URL: https://issues.apache.org/jira/browse/SOLR-11720
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: SolrCloud
> Affects Versions: 7.1
> Reporter: Vitaly Lavrov
>
> Scenario:
> 1. Run backup restore as async command
> 2. Check command status until it has 'completed' status
> 3. Query collection immediately for total number of documents
> The problem is the query will return incorrect number. The query will return
> correct number eventually if one keeps querying.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
