In article <aojkjf$6eh$1@sunsite.dk>, "Bruun" <k.bruun@agfaservice.dk>
wrote:
> Hi,
>
> We have encountered some problems regarding ufsrestore.
>
> I have posted the script we are using to dump a system hdd to another hdd.
>
> #!/bin/sh
> # script til system-disk backup
>
> echo "##############################################"
> >>/usr/local/agfa/sysdump.log
> echo `date` >> /usr/local/agfa/sysdump.log
> newfs /dev/rdsk/c0t2d0s0
> echo "NEWFS ON c0t2d0s0 DONE" >> /usr/local/agfa/sysdump.log
> sleep 5
> mount /sys_bak
> echo "MOUNT SYS_BAK" >> /usr/local/agfa/sysdump.log
> grep /sys_bak /etc/mnttab >> /usr/local/agfa/sysdump.log
> echo "GREP DONE" >> /usr/local/agfa/sysdump.log
> sleep 5
> ufsdump 0f - /dev/rdsk/c0t0d0s0 | (cd /sys_bak;ufsrestore xf -)
> echo "SYS DUMP DONE" >> /usr/local/agfa/sysdump.log
> sleep 5
> installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk /dev/rdsk/c0t2d0s0
> sleep 5
> umount /sys_bak
> echo `date` >> /usr/local/agfa/sysdump.log
> echo "##############################################"
> >>/usr/local/agfa/sysdump.log
>
> When we are running this script through crontab, it sometimes hangs with the
> ufsrestore command in memory. This results, that /tmp, /var/run and swap is
> filled and the server is forced in its knees.
>
> I have looked through /var/adm/messages in order to find the problem. The
> only message I get is this :
>
> Output from /var/adm/messages :
>
> Oct 16 08:16:38 Agfa-Server tmpfs: [ID 518458 kern.warning] WARNING: /tmp:
> File system full, swap space limit exceeded
> Oct 16 08:16:41 Agfa-Server last message repeated 17676 times
> Oct 16 08:16:43 Agfa-Server syslogd: malloc failed: dropping message:
> Resource temporarily unavailable
> Oct 16 08:16:41 Agfa-Server tmpfs: [ID 518458 kern.warning] WARNING: /tmp:
> File system full, swap space limit exceeded
> Oct 16 08:16:43 Agfa-Server syslogd: malloc failed: dropping message:
> Resource temporarily unavailable
>
> I have installed the latest recommeded patches and have looked through
> sunsolve.sun.com to find a possible patch. All the patches I could find are
> included in the recommended patches.
>
> We have encountered the same problem on Solaris 2.6 og 8
>
> When I run the script manually, it runs without ufsrestore hanging.
>
> Any bright ideas? I am totally lost? There shouldn't be any difference
> between running it through crontab og manually??
>
> I have contacted Sun Denmark for this issue. Where lucky enough to meet a
> incompetent helpdesk supporter. This sucks bigtime.
>
> Any one in here, who might have a solution?
Is /tmp the tmpfs filesystem? The log file says that swap space is
exceeded. Perhaps your cron job uses /tmp to create files by default,
even if it doesn't explicitly.
Try making /tmp a real filesystem.
--
DeeDee, don't press that button! DeeDee! NO! Dee...