Insufficient dynamic memory

[Sk Noorul Hassan]

I have got VAX 4105 A/VMS 7.1 platform, where a SCADA application is running which gets the data through Front End Processors(FEP) and in turn FEP scans the out stations(Remote Terminal Units) to get the real time data.

FOR this lots of FEP processes are running on my system. Sometimes we get 'Insufficient Dynamic Memory' messages and some FEP process are shown suspended in the system and that led to system crash.

Pls suggest the solution.. We are yet to implement the suggestion given by Jan on WSMAX sysgen parameter. Is this problem is related to any perticulat sysgen parameter ?

[Kris Clippeleyr]

Hi,

What is the exact mnemonic of the errors you see? Is it INSFMEM ? If so, you should check for programming errors since the FEP (probably) have exhausted the paged pool. OTOH, you might want to increase the paged pool (SYSGEN parameter PAGEDYN).

Regards,
Kris (aka Qkcl)

[Ian Miller.]

when this is seen look at
SHOW MEM
and
SHOW MEM/POOL

for lack of pool and free memory

[Volker Halle · Accepted solution]

Hi,

INSFMEM - if that is what you're seeing - is used in different parts of the operating system to indicate memory/pool (or similar) resource shortage. It depends on the actual system service or routine called, which resource really is exhausted.

You can check for paged or nonpaged pool allocation failures in the running OpenVMS VAX system or in the system dump file:

$ ANAL/SYS or ANAL/CRASH SYS$SYSTEM:SYSDUMP.DMP
SDA> SHOW SYMB/ALL pms$gl_npag
SDA> SHOW SYMB/ALL pms$gl_pag

all counters ending with 'F' indicate Failures.

If you have seen a system crash, consider to provide the CLUE file (from CLUE$OUTPUT:CLUE$LAST_node.LIS) as an attachment.

Volker.

[Hein van den Heuvel]

Noorul Hassan,

Please consider assigning points to the replies you have received to give those that have helped you some feedback as to how usefull their efforts were.

This will also help future readers to select the most critical replies to read.

Check out the FAQ:

http://forums1.itrc.hp.com/service/forums/helptips.do?#34

You can practice by assiging 0 points to this reply please.

Regards,
Hein

[Sk Noorul Hassan]

Thanks..

Can you, pls suggest if there is any way to monitor the system resources, to avoid system crash due to 'Insufficient Dynamic Memory'.

[Robert Gezelter]

Noorul Hassan,

As several posters have mentioned, insufficient dynamic memory can have many causes. For a start, the SHOW MEMORY command will show the gross amounts of dynamic memory used.

To identify exactly what is using the memory, it may be necessary to analyze a crash dump of the system.

I hope that the above is helpful.

- Bob Gezelter, http://www.rlgsc.com

[Volker Halle]

Hi,

you've never showed us the CLUE file from your crash. The PMS$GL_* pool statistics referenced earlier are available in the CLUE file (or the dump). If you believe the crash has happened to due 'Insufficient Dynamic Memory' - wherever you've seen this message (please show us), then you can verify from the CLUE file or dump, whether there have been pool allocation/expansion failures...

If there is some pool leak, so that the system is continously consuming more pool, you could see this with SHOW MEM/POOL/FULL:

- watch current size against initial size for nonpaged pool. If current size is increasing over initial size, it's time to do an AUTOGEN with feedback. If current size gets near maximum size, it's time for a reboot, as nonpaged pool cannot expand beyond NPAGEVIR.

- for PAGEDYN, watch Free Space.

If the error happens due to some software problem in a privileged application, so that it suddenly starts consuming lots of pool, monitoring pool usage may not help, as it's too late to do something.

As I've often said, OpenVMS is very good at providing tools for analyzing problems - even after they've happened. One of those tools if the crashdump and SDA, you just have to make good use of those tools. We are here to help you...

Volker.

[Sk Noorul Hassan]

Hi,
Thanks for your suggestion. This time after crash I will definitely attach the clue file. Insufficient dynamic memory error comes in application process warning logs.

In my case, Initial npagedyn and current size is equal in all the servers and npagevir is almost 5 times than npagedyn.

Pls suggest

[Volker Halle]

Hi,

so currently, you do NOT have a problem with the expansion of nonpaged pool. What's the current size of Free space in relation to Current size ? Maybe you could post the output of SHOW MEM/POOL/FULL from your system ?

Do you currently see error messages (insufficient Dynamic Memory) from your application ?

A CLUE file is automatically created after a system crash, so if you have a crashed system, which has not been shut down afterwards, you'll still have a CLUE$OUTPUT:CLUE$LAST_node.LIS file from the crash. You can also extract information about the (up to 100) previous crashes and shutdowns from the CLUE$HISTORY.DATA file.

$ clue:==$CLUE
$ CLUE/DISPLAY
CLUE> DIR ! will show your crash/shutdown history
CLUE> SHOW SYMB n ! n = crash number
will show the PMS$GL* symbols from each crash (or shutdown), so you can easily find out, if there have been pool expansion problems in your previous crashes (or shutdowns). If you find a crash, which you believe is related to INSFMEM and you find no pool allocation failures shown in the dump, you might have to re-think your analysis !

Could you post the real application error message ? Could it just be a UCX related problem ? Or even an application problem ?

Volker.

[Antoniov.]

If your virtual memory is enough large may be your user has limited page file quota. Look at authoiration file.

Antonio Vigliotti