Mailinglist Archive: opensuse-security (232 mails)

< Previous Next >
RE: [suse-security] Strange server crash
  • From: trainier@xxxxxxxxxx
  • Date: Thu, 13 Oct 2005 10:27:09 -0400
  • Message-id: <OF8865EB8E.FD168132-ON85257099.004F0604-85257099.004FBA23@xxxxxxxxxxxxxxx>
> I have tried tar, rsyn, and rdiff backup all of which have had similar
> memory leak problems causing kernel panics.
> Tar caused the problem when writing many files to a single large
archive.
> Rsync and rdiff-backup caused problems with the high volume of files.

High volume of files with rsync?
That doesn't make any sense. Rsync is designed to compare the two systems
to decide what files have
changed, as opposed to everything.

Also, does reiserfs not have a filesystem dumping utility like the
`dump/restore` utilities for ext2/3 ?

That is really what you should be using, if one exists. If not, I would
get ahold of the developer folks
at Novell and ask them why they haven't developed one yet. :-)

Unfortunately, I fear as though we've wandered off the path of on-topic
discussions for this list.
I certainly don't mind it, but others might. What other suse mailing
lists are out there?

Tim Rainier
Information Services, Kalsec, INC
trainier@xxxxxxxxxx

"b@rry" <b@xxxxxxxxx> wrote on 10/13/2005 06:11:32 AM:

> > I looked over the log files in the last 3 crashes ( 12 oct, 9 sept, 1
> sept)
> > and I couldn't asociate it with any cron or script or process.... only
if
> > one of my script hangs... that is again strange, because I do only
copy,
> database dumps...
>
> > I also do rsync with another share from network.
>
> I have tried tar, rsyn, and rdiff backup all of which have had similar
> memory leak problems causing kernel panics.
> Tar caused the problem when writing many files to a single large
archive.
> Rsync and rdiff-backup caused problems with the high volume of files.
>
> I have now resorted to "batching" my backups with date stamps inserted
into
> log files so that I can trace any further errors to a direct script and
> time.
> When backing up using rsynch, try inserting date stamps one dir at a
time,
> that way at least you will know if a perticular dir/process is causing
the
> problem.
>
> Bit of a pain in the ass, but less of a pain than a panic.
>
> The dir structure I backup is basically:
> /top/of/dirstructure/0/0/0/0/0
> To
> /top/of.dir/structure/9/9/9/9/9
>
> With each dir containing
> 0/0/0/0/0
> 0/0/0/0/1
> 0/0/0/0/2
> 0/0/0/0/...9
> 0/0/0/1/0
> 0/0/0/1/1
> 0/0/0/1/2
> Etc
>
> So large number of dirs.
>
> I have scripted
>
> #!/bin/bash
>
> rm -rf /var/log/backup.log
> echo "Start Dir Struc 0" `date` /var/log/backup.log
> Rdiff/rsync /top/of/dirstructure/0/
> echo "End Dir Struc 0" `date` /var/log/backup.log
> sleep 2
> echo "Start Dir Struc 1" `date` /var/log/backup.log
> Rdiff/rsync /top/of/dirstructure/1/
> echo "End Dir Struc 1" `date` /var/log/backup.log
> sleep 2
>
> Etc...
>
> This way, I was able to ensure that if anything went wrong I would
always
> know very closely where and could check my dir contents for file
allocations
> that were haywire.
>
> Also, breaking it up into many processes while it makes the backup
process
> take much longer, I have not had a single issue since doing this...

< Previous Next >
Follow Ups
References