[EMAIL PROTECTED] writes:
> I'm currently using backuppc in a few environments, with rsyncd as
> the primary protocol. I'm running into issues of high memory use on the
> client, and seeing that full backups are taking a very long time. Backups
> often fail due to timeouts/memory use on the client or need to be killed
> because the backup window runs into "peak time" and the load due to
> rsync is affecting production use of the servers.
>
> My goal is to reduce the impact on clients of the BackupPC server during full
> backups, and to make the backup window smaller. Reaching these aims will
> require more resources (particularly disk space) and more processing time on
> the backup server, and may incur a higher network load than doing backups as
> they are now.
>
> I understand the benefits of using rsync for incremental backups, but not for
> full backups. My change would be to get a full backup of the client data to
> the
> backup server as quickly as possible, then let the backup server do the
> "rsync"
> locally, in order to get the advantages of pooling and compression.
>
> I'd like to propose an enhancement to BackupPC to speed up full backups:
>
> allow the specification of an "external" mechanism for doing a full
> backup from a client to the BackupPC server. This could be a dynamic
> "plugin" or even a manual process.
>
> As I envision it, the configuration would specify:
>
> the name of the client
>
> the location where the full backup will be temporarily stored
> on the BackupPC server
>
> commands to run on the client and server to transfer the data
> (this could simply be "manual")
>
> Once the data is on the BackupPC server, the server would then treat
> that directory tree as if it was the source of a backup, and do a
> local backup from that tree to the $BACKUPPC/pc/$CLIENT directory,
> building NewFileList and $BACKUPPC/pc/$CLIENT/new as if it was
> receiving the data from the network, including doing all compression
> and pooling as usual.
>
> For example, I could specify that the data be transferred by running "dump" on
> the client and "restore" on the backup server (a much more efficient process
> than rsync). Alternatively, the full backups could be performed by breaking a
> mirrored drive on the client, attaching the disk to the server as an external
> device, mounting the filesystems, and using them as the source. The mechanism
> for moving the data would be external to BackupPC.
>
> So, what do you think?
This sounds reasonable. It should be possible to do this
by making $Conf{RsyncClientCmd} point to a script that you
write. It could do the copy, and then run rsync locally
on the data.
Also, Roy Keene is working on a BackupPC client that will
avoid many of the issues that rsync has.
Craig
-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. Download
it for free - -and be entered to win a 42" plasma tv or your very own
Sony(tm)PSP. Click here to play: http://sourceforge.net/geronimo.php
_______________________________________________
BackupPC-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/backuppc-users
http://backuppc.sourceforge.net/
