[prev in list] [next in list] [prev in thread] [next in thread] 

List:       forgerock-opendj
Subject:    Re: [Opendj] backup / restore
From:       Ho Tri Bao <hotribao () gmail ! com>
Date:       2014-04-27 14:28:58
Message-ID: 535D142A.90307 () gmail ! com
[Download RAW message or body]

Thank you very much Ludovic for the concise response.

For the text format, I was looking good in my Thunderbird :D I'll take care of it

Regards,
Bao.


On 04/27/2014 06:00 PM, Ludovic Poitou wrote:
> Hi Bao,
> 
> (Just a side note : you email would be much easier to read if it was in plain text \
> and not html with huge fonts). 
> 1/ Incremental backup behaviour description is correct.
> 
> 2/ Restore is meant to restore a backend, and thus you need to provide the backend \
> you want to restore. Most of the time, you're concerned about the data, and not the \
> schema or directory topology data (cn=admin data). It is a best practice to restore \
> a backend one at a time. Task is a special backend that holds current executed or \
> scheduled tasks such as backups. When you want to run a backup online, the backup \
> command creates over LDAP a Task entry, which is executed by the server. When the \
> task is finished, the server removes the object (after a short delay). Likewise, \
> schema is a special backend that holds all of the schema definitions and files. 
> 3/ Restore without a backupID will restore the latest backup. If incremental is \
> used, it'll restore the latest full backup and all incremental ones on top of it. \
> Replication initialisation is meant to ensure the data on a new replica (or an \
> existing one when reinitialising) is the same as one the other servers. For this \
> reason, replication initialisation does always do a full transfer of the data. 
> Regards,
> 
> Ludovic.
> 
> --
> Ludovic Poitou
> Product Manager
> ForgeRock
> 
> On 27 Apr 2014 at 11:36:32, Ho Tri Bao \
> (hotribao@gmail.com(mailto:hotribao@gmail.com)) wrote: 
> > Hi,
> > 
> > I'm new to OpenDj. I'm trying to setup a backup/restore strategy for our OpenDJ \
> > cluster. 
> > I read http://opendj.forgerock.org/opendj-server/doc/admin-guide/index/chap-backup-restore.html \
> > and I am able to do backup and restore on my local OpenDJ instance. However, \
> > there are somethings that I would like to confirm: 
> > - Is this behavior correct: When I run the "backup" command with the \
> > "--incremental" but without an "--incrementalBaseID": + If the backup directory \
> > is empty, a full backup will be created. + If there is a backup (either full or \
> > incremental), a new incremental backup will be made for data since the last \
> > backup store in that directory. I would like to confirm this behavior because I \
> > want to make daily incremental backup + one weekly full backup by just using one \
> > command "backup ... -incremental" (without --incrementalBaseID) and control the \
> > backup directory: when wanting a full backup, just provide the script an empty \
> > directory. 
> > - Below are questions regarding to the restoration process:
> > + If I give the directory "/backup" to the "backup" command and also give that \
> > directory to the "restore" command, the "restore" command will refuse to work \
> > with a message "No backup directory descriptor file was found at \
> > /backup/backup.info". The file "backup.info" appears in 3 different location: \
> > /backup/schema, /backup/tasks and /backup/userRoot. I give the "/backup/userRoot" \
> > to the "restore" command, it seems to work OK. The questions are: (?) providing \
> > "/backup/userRoot" as "backupDirectory" to the "restore" command, is it correct? \
> > (?) is it correct that, if we have many backends, for each backend, we need to \
> > separately run a "restore" command? (?) what are the "schema" and "tasks" \
> > subfolder used for? 
> > + Is it correct that, if I don't privide "backupID" to the "restore" command, it \
> > will restore everything (upto the last backup) in side the backupDirectory? 
> > - Assume I have a cluster of 2 nodes. Now I need to add one more replica. If I do \
> > as following: + I install a new node from scratch.
> > + I restore the last backup to the new instance.
> > + I perform replica initialization.
> > Will this help reduce the initialization? (as I assume the initialization process \
> > will only transfer a delta data to the new node) 
> > Kind regards,
> > Bao.
> > _______________________________________________
> > OpenDJ mailing list
> > OpenDJ@forgerock.org
> > https://lists.forgerock.org/mailman/listinfo/opendj

_______________________________________________
OpenDJ mailing list
OpenDJ@forgerock.org
https://lists.forgerock.org/mailman/listinfo/opendj


[prev in list] [next in list] [prev in thread] [next in thread]