Error Message

[Robie Lutsey]

I have a user that gets the message below. She had about 12 Hpterms open. Could that be part of the problem? It's a J5000 wth HPUX 11.0.

/usr/lib/dld.sl: Call to mmap() failed - BSS /esri/arcexe72/lib/libelib.
sl
/usr/lib/dld.sl: Not enough space
Bailing out of ARCEDIT

Program failed to start or was killed

Try RESTOREARCEDIT to recover your edit session
FATAL ERROR
Invalid LUN (KLOC)

[A. Clay Stephenson]

This could be caused by lack of swapspace or maxdsiz set too low. Having many processes going could begin eating swap so that too could cause it.

[Sridhar Bhaskarla]

Did you check up your swap space utilization during this time?. This is basically because the kernel is unable to allocate the addresses in VAS which may also be due to maxdsiz kernel parameter limitation.

-Sri

[S.K. Chan]

Usually it means your system is running out of process space trying to map shared library data into the BSS area. The most common solution to this error is to increase the swap.

[Robie Lutsey]

Can increase the current swap or do I have to Add another swap?

It looks like SAM won't let me increase the current one.

[James R. Ferguson · Accepted solution]

Hi Clifford:

It is very unlikely that you will be able to extend your primary swap since contiguous extents are required. This is not a problem, however.

You can add more device swap without having to reboot (assuming that you have a sufficient value for 'maxswapchunks' already defined for your kernel.

Simply create a secondary device swap on a disk other than one in your vg00. Make the size the same as your primary swap space and assign an equal priority. In this fashion you will get I/O interleaving and the best performance.

Regards!

...JRF...

[Sanjay_6]

Hi Clifford,

Try this link below,

http://us-support2.external.hp.com/cki/bin/doc.pl/sid=4cb41631039055af1c/screen=ckiDisplayDocument?docId=200000050215839

Hope this helps.

Regds

[Robie Lutsey]

Thanks everyone!

I added more swap space and tweaked the kernel. Reboot tommorrow! Thanks!!

[Bill Hassell]

Make sure that swapspace is the issue by running swapinfo -tm and see if the first line totals for device swap (when the machine is busiest or experienceing the error) is pushing more than 80-90% occupied. If so, you can temporarily add extra space by using the swapon command and pointing it to a mounted filesystem. This can get you over the problem temporarily.

On the other hand, if the process that is being aborted is pushing more than 50 megs in size, the kernel parameter maxdsiz (which protects from runaway programs that consume all RAM) may be set to the default of 64 megs. Use SAM to bump this up to 200 or 400 megs...it does not affect anything as far as running, just sets a limit on how big a program can grow. Use this command to sort all the processes by size:

$ UNIX95= ps -e -o vsz,pid,args | sort -rn | more

The first column is Kbytes for the process. Anything in the 40000-60000 range is pushing the default limit for maximum process size (maxdsiz).