Memory limitations on 11iV1 system

[Stuart Powell]

I have two development systems the mirror our production environment running HP-UX 11iV1 and ServiceGuard 11.16. We are encountering some problems running backups via Data Protector 6.0 on our test Informix 11.50 database. We receive the following error message:

[Critical] From: OB2BAR@th9ktux.aetinc.net "dwtcp" Time: 3/19/2009 3:24:18 PM
[80:1011] Cannot fork NETIO process ([12] Not enough space) => aborting.

[Critical] From: OB2BAR@th9ktux.aetinc.net "dwtcp" Time: 3/19/2009 3:24:18 PM
15
[Critical] From: OB2BAR@th9ktux.aetinc.net "dwtcp" Time: 3/19/2009 3:24:18 PM
15
We have been running near 100% on memory and 95% on swap space for months. It seems to me that even though memory is a contributing factor it is not THE change that created the problem.
I'm thinking that I may need to increase both RAM and device swap sizes. But I'd like to know if anyone is aware of changes in Informix 11.50 that deal with memory differently than 10.X. Or maybe it is a difference with the Data Protector Informix integration.
I do have a call open on this problem, but wanted to get any feedback available from the user community.

[Steven E. Protter · Accepted solution]

Shalom,

I would like to see swapinfo -tam information.

A system running at near 100% memory is not a problem. Neither is reservation of swap.

Check vmstat to see if actual paging is going on. Also, if there are no user response complaints, you may wish to do nothing.

Also here is a memory leak detector utility that might help you identify processes that show growth in memory use.

http://www.hpux.ws/?p=8

SEP

[RUET]

It's sure that it's a virtual memory problem. You can have a RAM USAGE closed to 100%, it's not a problem if you have free space in swap area (system will just be slow because of swapping activity).
You should increase swap space, taht can be done online.
If memory is still used at this ration, next will analyse which components is responsible of this, and if no solution can be found because your informix database is taking memory, you will have to increase RAM

patrick

[T G Manikandan]

Error 12 (ENOMEM) - clearly states that its "not enough memory".

you need to check the utilization of swap on the system, swapinfo -tam and also the current memory and swap size on the system.

[Stuart Powell]

Steven,
Here is the swap information:
$ swapinfo -tam
Mb Mb Mb PCT START/ Mb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 4096 181 3915 4% 0 - 1 /dev/vg00/lvol2
reserve - 3486 -3486
memory 4624 4624 0 100%
total 8720 8291 429 95% - 0 -

I have been watching page out in glance and finally saw some page out activity, but for the most part it has been quiet. I only saw that one burst.

[Dennis Handly]

>but for the most part it has been quiet.

Then you need more device swap. Or remove some of the load on your machine.