lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Della Bitta <michael.della.bi...@appinions.com>
Subject Re: How to Recovery Backup Snapshot at SolrCloud?
Date Wed, 01 May 2013 14:26:29 GMT
Yeah, it's a consistency problem. Copying all those files takes time, and
without something with some knowledge of how Lucene works managing the
atomicity of the work, you might end up with a segments file that doesn't
match the segments you actually copied.


Michael Della Bitta

------------------------------------------------
Appinions
18 East 41st Street, 2nd Floor
New York, NY 10017-6271

www.appinions.com

Where Influence Isn’t a Game


On Wed, May 1, 2013 at 8:59 AM, Furkan KAMACI <furkankamaci@gmail.com>wrote:

> Sorry but what will I do? Will I copy everything under snapshot folder into
> under index folder? If I don't run backup command and just copy index
> folder anywhere else what is the difference between them (is it something
> like consistency for if any writing operation for segment files did not
> finished, or does backup command retrieves indexes data at RAM)
>
>
>
> 2013/5/1 Timothy Potter <thelabdude@gmail.com>
>
> > I agree with Michael that you'll only ever need your backup if you
> > lose all nodes hosting a shard (leader + all other replicas), so the
> > tlog doesn't really factor in when recovering from backup.
> >
> > The snapshot created by the replication handler is the index only and
> > it makes most sense in my mind to remove the tlog before firing up a
> > new node using the snapshot.
> >
> > The way I see it, the backup would be needed if you lost all nodes
> > hosting a shard, then you need to 1) recover one node (the leader)
> > from backup, 2) re-index any documents that were indexed between the
> > time of your last snapshot and the time of the failure, then 3) bring
> > additional replicas online as needed. The additional replicas will
> > sync with the leader, snap pulling the entire index. You could do it
> > 1,2,3 or 1,3,2.
> >
> > Cheers,
> > Tim
> >
> > On Tue, Apr 30, 2013 at 12:45 PM, Furkan KAMACI <furkankamaci@gmail.com>
> > wrote:
> > > I had index and tlog folder under my data folder. I have a snapshot
> > folder
> > > too when I make backup. However what will I do next if I want to use
> > > backup, will I remove index and tlog folders and put just my snapshot
> > > folder? What folks do?
> > >
> > > 2013/4/30 Michael Della Bitta <michael.della.bitta@appinions.com>
> > >
> > >> Presumably you'd only be restoring a backup in the face of a
> > catastrophe.
> > >>
> > >> Yes, you'd need to stop the node. And the transaction logs may not be
> > >> useful in this case. You'd have trouble reconciling them with the
> > version
> > >> of the index in your backup I would think.
> > >>
> > >> Anybody who knows more about this want to chime in?
> > >>
> > >>
> > >> Michael Della Bitta
> > >>
> > >> ------------------------------------------------
> > >> Appinions
> > >> 18 East 41st Street, 2nd Floor
> > >> New York, NY 10017-6271
> > >>
> > >> www.appinions.com
> > >>
> > >> Where Influence Isn’t a Game
> > >>
> > >>
> > >> On Tue, Apr 30, 2013 at 11:03 AM, Furkan KAMACI <
> furkankamaci@gmail.com
> > >> >wrote:
> > >>
> > >> > Should I stop the node first? And what will happen to transaction
> > logs?
> > >> > Should I backup it too?
> > >> >
> > >> > 2013/4/30 Michael Della Bitta <michael.della.bitta@appinions.com>
> > >> >
> > >> > > That directory is the data directory for the core... you'd just
> > swap it
> > >> > in.
> > >> > >
> > >> > >
> > >> > > Michael Della Bitta
> > >> > >
> > >> > > ------------------------------------------------
> > >> > > Appinions
> > >> > > 18 East 41st Street, 2nd Floor
> > >> > > New York, NY 10017-6271
> > >> > >
> > >> > > www.appinions.com
> > >> > >
> > >> > > Where Influence Isn’t a Game
> > >> > >
> > >> > >
> > >> > > On Tue, Apr 30, 2013 at 8:06 AM, Furkan KAMACI <
> > furkankamaci@gmail.com
> > >> > > >wrote:
> > >> > >
> > >> > > > Hi Folks;
> > >> > > >
> > >> > > > I can backup my indexes at SolrCloud via
> > >> > > > http://_master_host_:_port_/solr/replication?command=backup
> > >> > > > and it creates a file called snapshot. I know that I should
pull
> > that
> > >> > > > directory any other safe place (a backup store) However
what
> > should I
> > >> > do
> > >> > > to
> > >> > > > make a recovery from that backup file?
> > >> > > >
> > >> > >
> > >> >
> > >>
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message