lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Maxim Novikov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SOLR-5946) Need clear docs on how to "backup" and "restore" indexes
Date Wed, 02 Apr 2014 02:11:15 GMT

    [ https://issues.apache.org/jira/browse/SOLR-5946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13957257#comment-13957257
] 

Maxim Novikov commented on SOLR-5946:
-------------------------------------


Please describe also the strategy for restoring backups, specifically for SorlCloud. From
your documentation it is still possible to grasp the idea how to back up the index using HTTP
commands, but it is absolutely unclear what to do with those backups if the cluster goes down
one day for some reason. So far those are just a bunch of files/directories no one knows how
to use.

Would we need to delete the directory with the current index?
Should we copy one  of the snapshot directories files to the index directory?
Would we have to do something with transaction logs (delete, modify, etc.)? 
How to let the other nodes in the cluster know that the index has been restored from an earlier
version and trigger the synchronization process?

Also, it does not seem to be explained anywhere, but it looks like fetching data (triggering
that process via HTTP) from master to slave does not work in SolrCloud as all the nodes are
considered to me masters.

I believe all these questions (and the others related) should be addressed as for now SolrCloud
is not being a comprehensive solution because you need to work around some similar stuff on
your own. And it is not really reliable if Solr cannot handle such things at all.

PS My understanding is that currently backups are useless as you cannot do anything with them
(at least I have not found any info that would cover that). They may help only in case of
complete horrible failure when all the nodes in the cluster got exploded and then you want
to restore the index quickly on other ones avoiding the full import from the data source from
scratch. But this is really not a real-life use case. The more typical would be "I want to
restore my data after I occasionally cleaned up the index in SolrCloud" or "I want to restore
my data from a backed up snapshot when the index got corrupted for some reason", etc.


> Need clear docs on how to "backup" and "restore" indexes
> --------------------------------------------------------
>
>                 Key: SOLR-5946
>                 URL: https://issues.apache.org/jira/browse/SOLR-5946
>             Project: Solr
>          Issue Type: Bug
>            Reporter: Hoss Man
>
> As pointed out by Maxim Novikov in the comments of the Solr Ref Guide: the instructions
on creating index backups are very sparse, and give no clear instructions on how to _restore_
fro ma backup.
> This page really needs to be beefed up with more complete instructions...
> https://cwiki.apache.org/confluence/display/solr/Backing+Up



--
This message was sent by Atlassian JIRA
(v6.2#6252)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message