Memory problems ...

[Dr. Peer-Joachim Koch]

Hi,

we are running a L-class with 2 CPU, 1GB, 2FC-HBA and 1 GB-Lan as a file server for our institute. The machine is running under 11.0. NFS and SAMBA are the only applications/services running on the machine.
A few weeks ago the TSM-Backup has start to cause problems. It dies with an error message "running out of memory". But we have change anything for the last 12 month and the backup was always running fine ...
The Backup by hand works, but is much slower.
Now I noticed also that the vi shows some
similar error message, when I tried to delete
some lines from a log-File:
Out of register space (ugh)
The file is just 17MB. So the system has a problem with the memory, but what can be the reason for it ?
If I just restart the server I can not be sure
to solve the problem.
The only part which changed was the network (we have had problems with our core switch). So the network link was sometimes missing.
Any idea ?

Thanks, Peer

[Chris Wilshaw]

The error from vi doesn't relate directly to your general memory errors. There are internal buffer limits to vi (when you yank or delete large blocks of a file) From memory, I think this is roughly 50000 lines, although it would also depend on the line lengths.

The only way to diagnose the memory problems during the backup is to use diagnostics (glance, vmstat etc) while the backup is running to see what in particular is taking up memory.

It may be that you can get away with just restarting the application processes on the server, rather than doing a full reboot - if this proves to be the case, you need to speak to the application vendor to see if they have any memory leak patches that you can apply.

[Stefan Farrelly]

Sounds like a slow memory leak to me too. What does swapinfo -mt show ? (swap usage) and vmstat (free memory) ?

A reboot will certainly fix it, and if its been running fine for many months then I wouldnt be too worried about it. The latest NFS megapatch and Samba patch/update may fix it.

[Jakes Louw]

Check for any defunct processes, and kill where possible. Also, check IPCS for any shared-memory segments that shouldn't be there.
Of course, a simple reboot will clear your problem until you can determine what the memory hogs are.....

[Paddy_1]

The problem is with vi.
It cant handle certain long lines or big files.Upgrade it to the latest version or use other vi clones which can handle such big files.

[Steven E. Protter]

Advice:

June 2003 or September 2003 Gold or Quality Pack installation.

Staying current is a very GOOD thing.

itrc.hp.com
indidvidual patches
hpux

search for vi patches

Install them.

I'd be interested to see if there are related messages in /var/adm/syslog/syslog.log

SEP

[Dr. Peer-Joachim Koch]

Hi,

I tested the vi with the same file on a similar machine with the same result. OK, this is a bug (or whatever) of the vi.

But the TSM system is somehow different. It work for 3 years without any problem. OK, we have now more diskspace (~4 TB, 7 million files), but this is not a big deal for tsm.

bash-2.03# swapinfo -mt
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 1024 0 1024 0% 0 - 1 /dev/vg00/lvol2
dev 1536 644 892 42% 0 - 0 /dev/vg00/lvol9
reserve - 528 -528
memory 905 905 0 100%
total 3465 2077 1388 60% - 0 -
bash-2.03# vmstat
procs memory page faults cpu
r b w avm free re at pi po fr de sr in sy cs us sy id
0 0 0 125173 6348 1 0 0 0 0 0 1 1519 2038 562 3 6 91

I will restart the machine the next time we have a maintenance day.

Thank's and bye, Peer

[Dr. Peer-Joachim Koch]

Hi,

there is NO entry in the syslog .

Bye, Peer

[Suresh Patoria]

Hi,

pls check the how much free space in /tmp and /var
using the bdf commnad

and to take backup which command using
pls reply this

Thanx