insufficient swap or memory available

[WSS]

Hi All, Urgent issue, this keeps scrolling up on the screen, and can do very limited things. Can't run glance or top.

Any points as to how to get the box back to running order? See output of swapinfo to see it's at 100%

Thanks....

root@udox305# swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4096 1584 2507 39% 0 - 1 /dev/vg00/lvol2
dev 4096 1577 2514 39% 0 - 1 /dev/vg00/lvol11
dev 4096 1574 2517 38% 0 - 1 /dev/vg00/lvol12
dev 12000 1583 10406 13% 0 - 1 /dev/vg00/lvol13
dev 4096 1559 2532 38% 0 - 1 /dev/vg00/lvol14
dev 4096 1584 2507 39% 0 - 1 /dev/vg00/lvol15
dev 4096 1583 2508 39% 0 - 1 /dev/vg00/lvol16
reserve - 25461 -25461
total 36576 36505 30 100% - 0 -
root@udox305# exec(2): insufficient swap or memory available.

[DCE]

Can you shut down any programs/applications/monitoring tools? i.e. measureware (or any other listings in a ps -ef} This should give enough resources so you can run glance and or top to determine if there is a runaway process using the resources.

[Peter Godron]

WSS,
release some memory by stopping processes.
For full details on swapinfo meaning:
ftp://hprc.external.hp.com/memory.htm#the_memory_line_in_the_output_of_swapinfo

[Mridul Shrivastava]

I see that reserve is accounting for all of the memory being used, with dev PCT
USED showing 0%.

if you start several instances of the program, then should be able to reserve virtually all of the memory on the system.Please check if there are multiple instances of a program running...

[WSS]

Ok, we HAD 7 instances of an application running, which we have had running for literally months, and hasn't been an issue before.

I shutdown one of the instances, and it has given me back about 12% now, so a swapinfo -tam now gives a total of 88%, so the users are now able to use the server, and given me some breathing space for trying to find some answers. One thing that worries me, is this process when I do a top -


0 ? 24549 root 241 20 17228K 2080K run 92:42 99.31 99.13 HPUXChild

Showing 99% of CPU Utilization. This is what I get when doing a ps -fe | grep 24549

/opt/java1.3/jre/bin/HPUXChildWrapper 6 9 11 /usr/sbin/cmviewcl

I have had some issues with Service Guard over the past week or so, and currently Service Guard isn't even running... I may wanna kill this process??

Thanks in advance....

[WSS]

Well I rebooted the server last night which fixed any memory issues, or processes highly intensive of CPU. Everthing runing ok this morning...

Regards,

[cmac]

Peter,

I assume that the URL you gave to that memory document is only for people with a HP support agreement?

[Bill Hassell]

Just a mote about performance. swapinfo reports a *lot* of occupied swap space which indicates a severe lack of RAM. There are only two fixes: add a *lot* more RAM or reduce the number of processes running at the same time. As you've seen, getting information about a problem program is not dependable because RAM is so limited. And of course, paging out (swapping) really slows overall performance.