Server error - Memory allocator

[Chopsy]

I have started to see the following error messages in my log file -

Cache memory allocator out of available memory.

Short term memory allocator is out of memory.
1 attempts to get more memory failed.

It only seems to have started recently, and no major changes have been made to the system.

Any suggestions to what is causing it and the best remedy?

Thanks

 

[spi200]

Hi Chopsy

This is an easy one - You need to add memory to your server.
If you check your cache buffers in monitor they will be low ie less than 40% probably as low as 20%. Add plenty of RAM and you will get a performance improvement as well and your cache buffers will be 70% or better.
In the short term you may be able to unload any NLM's not critical after a clean reboot.

Regards

David

 

[Chopsy]

Under Memory Utilisation, it shows Percent Free 9%. Is this the figure you're talking about?

 

[spi200]

Hi Chopsy

Its under System Resources and its called CACHE BUFFER MEMORY - xx% where xx should be above 40%. I have my server at 79% but I have around a gig in the servers. My rule of thumb is double the amount of RAM if your server needs it, ie If you have 256Mb Upgrade 512Mb etc.

regards

David

 

[TheLad]

What memory do you have? That error is not always caused by a lack of memory as I have six servers configured with the same memory and diskspace and I have one coming up with that error at the moment. Will post my findings... -----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[MarkhP]

Can be caused by an out of date CLIB.NLM file. Also caused sometimes if reserved buffers below 16meg statement in startup.ncf is set high (try setting to 200). Cache buffers should be above 60%. Certainly anything below that, and your asking for trouble. Another thing to try is to purge deleted files, as too many cause problems.

 

[JockMullin]

Hi, Chopsy

I have also had this problem crop up unexpectedly. It turned out to be due to a large number of files having been allocated on the server's SYS volume. As a result when Novell sets up the directory cache it uses a whole whack more memory than previously.

Try dismounting a volume and see if that alleviates the problem (assuming you are not using SYS for everything).

It also means memory was tight in the first place. With today's memory prices, just throw more memory at it (the Windows solution).

Brgds
Jock

 

[seancane]

Chopsy and all that are having these problems. Cache memory allocator out of available memory is an age old "memory leak" error. 99% of the time it is the result of a misbehaved NLM (or .dsk, .lan, .ham) (or a conflict between NLM, HAM, DSK, LAN, etc).

Unless the system comes up with this error IMMEDIATELY on reboot, it is not the result of not enough memory, or too many files on a volume (although purging all volumes is a great way to use less memory).

You need to look at the amount of memory being taken by each NLM, and find the one that is trying to take more memory than it should need. If you have a similar server, then compare the lists. Normally, patching and updates will fix this problem. Throwing more memory at a memory leak issue just delays when the error comes back.

Be methodical. Create a baseline of how much the NLMs take, and then watch to see which NLM grows out of whack with the others.


Sean