Hi,
Thanks for your input - I further analyzed the issue and noticed the
following:
1) At time of the "hang" memory (only 128MB, will be more this week :-)
memory was full and swap was 19% in use -but-
2) In my "top" output I saw that dump, imap-login etc. where waiting with
status pfault
3) Due to lack of memory I had added an additional swap-file via mdconfig 3
days ago and due to space-restraints I had to put it into /usr (not a good
choice, but lacking space everywhere else) --> so I'm backing-up a
filesystem with snapshots active and the swapfile is in the same filesystem
I saw some discussion about issues with 6.0 and swapfiles in filesystems but
with no conclusions...
Thxs in advance, cheers,
-Martin-
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of backyard
Sent: Montag, 21. August 2006 15:55
To: Martin Werner; freebsd-questions@freebsd.org
Subject: Re: Hangs during "dump" with 6.0 and current ports
I've had problems with dump and restore on machines
lacking memory before. Perhaps the dump is just
running the system out of memory? I know I've had
issues restoring my /usr filesystem with 512M RAM
unless I had a swapfile active. I also find it helps
to make sure /tmp has got enough space on it at least
on restores.
Maybe you need some more swap space? Since its dumping
live filesystems and your running services then the
snapshot would have to be buffered somewhere, and in
your case swap would likely be where.
-brian
--- Martin Werner <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I'm running 6.0 on a Celeron 800MHz with 128MB and
> just updated the system
> to current ports (incl. perl 5.8.8).
>
> The systems is running apache, mysql, postfix and
> dovecot with a minor load
> (especially at the time when the system is
> backed-up)
>
> The issue I have is that my backup-script using
> "dump" will keep the system
> in a state where the IP-Stack is still there (can
> ping an scan ports), but
> no requests whatsoever are accepted and due to that
> I can't access the
> system via ssh and tell what's going on.
>
> The last time I had a ssh-session open with "top":
> Here is the first few
> lines output:
>
> snip --
> last pid: 948; load averages: 0.00, 0.00, 0.00
> up 0+00:37:51 08:54:13
> 72 processes: 1 running, 62 sleeping, 9 waiting
> CPU states: 0.0% user, 0.0% nice, 0.4% system,
> 0.4% interrupt, 99.2%
> idle
> Mem: 59M Active, 16M Inact, 39M Wired, 288K Cache,
> 22M Buf, 480K Free
> Swap: 357M Total, 69M Used, 288M Free, 19% Inuse
>
> PID USERNAME THR PRI NICE SIZERES STATE
> TIME WCPU COMMAND
> 591 mysql 5 200 56884K 1680K kserel
> 0:05 0.00% mysqld
> 658 root1 960 24344K 15948K select
> 0:04 0.00% perl5.8.8
> 908 root1 960 2416K 1436K RUN
> 0:03 0.00% top
> 722 root1 960 26132K 0K WAIT
> 0:02 0.00%
> - end --
>
> Here is the script I am running via cron:
>
> --- snip ---
> #!/bin/sh
> dump=/sbin/dump
> wput=/usr/local/bin/wput
> chflags=/bin/chflags
> dt=`date +%Y%m%d`
> destpath=/usr/dump
> logfile=$destpath/backup.log
> lvl=""
> lvl=$1
>
> if [ "x$lvl" != "x" ]
> then
> echo "Backup Level: " $lvl > $logfile
> else
> echo "No Backup-Level specified - exiting" >
> $logfile
> exit 911
> fi
>
> # /
> src1=/dev/ad0s1a
> # /var
> src2=/dev/ad0s1d
> # /usr
> src3=/dev/ad0s1f
>
> dest1=root_ad0s1a_$dt.gz
> dest2=var_ad0s1d_$dt.gz
> dest3=usr_ad0s1f_$dt.gz
>
> # Ausnahmen NO BACKUP
> $chflags -R nodump /usr/ports/ >> $logfile 2>&1
> $chflags -R nodump /usr/src/ >> $logfile 2>&1
> $chflags -R nodump /usr/obj/ >> $logfile 2>&1
> $chflags -R nodump /usr/dump/ >> $logfile 2>&1
> $chflags -R nodump /usr/swapfile2 >> $logfile 2>&1
>
> # Fullbackup Level 0 Monatlich
> $dump -$lvl -h 0 -Lauf - $src1 | gzip -2 | dd
> of=$destpath/$dest1 >>
> $logfile 2>&1
> $dump -$lvl -h 0 -Lauf - $src2 | gzip -2 | dd
> of=$destpath/$dest2 >>
> $logfile 2>&1
> $dump -$lvl -h 0 -Lauf - $src3 | gzip -2 | dd
> of=$destpath/$dest3 >>
> $logfile 2>&1
> end ---
>
> The "chflags" are executed correctly, the first two
> dumps are done and
> during the 3rd "dump" execution against the "/usr"
> Mount-Point