on 01-13-2015 6:51 AM
Hi experts,
I'm trying to apply NetWeaver 740 SP08 to IDES with SUM but it fails with shortdump repeatedly.
Now I'm on Checks step in SUM, PREP_GENCHECKS/JOB_RSUPGRCHECKPRE, and the job RSUPGRCHECK aborted with shortdump STACK_NO_ROLL_MEMORY.
Source SP level is 740 SP07, OS is Windows 2012, DB is HANA rev82.
The error occurs in "SAPLRSDG_IOBJ_DB_READ", in "GET_IOBJ_FROM_DB".
Virtual memory in operating system level is still available at the dump generation, abap/heap_area_total is not reached.
I have adjusted em/initial_size_MB, em/address_space_MB, and abap/heap_area_total but EM+Heap used sizes at termination were almost the same.
Does anyone have an idea to tune and resolve the issue?
Thanks in advance.
Megumi
I got fixed the issue with help of SAP development team support.
The issue was caused by remote hierarchy, then I applied following two notes.
2041767 - CPM: System dumps with Errors mentioning 'No Memory Available'
1990210 - CO-OM Planning: Program Termination with Runtime Error TSV_TNEW_OCCURS_NO_ROLL_MEMORY
After these correction, I wasn't required such a huge memory.
Thanks all.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I found the same message in these several shortdumps.
> You attempted to extend a stack, but the required storage space was
> not available.
> Storage request..............: 2650762512 bytes.
Now, my profile parameter "ztta/max_memreq_MB" is set to 2047 and this is the max value of this parameter.
It seems the requested size exceeds the limitation. Is it the cause?
Does anyone have an idea?
Thanks,
Megumi
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Megumi,
I think, you're looking at a wrong parameter. The ztta/max_memreq_MB parameter determines the maximum size of an individual memory request.
Please increase the em/initial_size_MB by 10% and try the step, provided you have enough memory.
You may then revert after the activity.
Regards,
Prithviraj.
Hi Prithviraj,
I have increased em/initial_size_MB several times and reached to 128,000, it is almost 3 times larger than initial value.
But ztta/roll_extention limits EM per user context and its maximum value is 64,000,000,000.
In my case, EM was fully used, about 64GB, and about 155GB HEAP was assigned.
abap/heap_area_total was 256,000,000,000 and for nondia was 0, that meant HEAP didn't exhausted.
That's why I doubt the relation between requested storage size and ztta/max_memreq_MB.
What do you think?
Regards,
Megumi
Hi Megumi,
Could you please paste the values for below memory parameters?
abap/heap_area_total
abap/heap_area_dia
abap/heap_area_nondia
abap/heap_limit
ztta/roll_extention
em/initial_size_MB
According to dump file you provided the process has used..
Extended memory (EM) 51904302708 = 51 GB
Assigned memory (HEAP) 133949209168 = 133 GB
The heap was not used completely (as you have set the nondia parameter to zero) means some other process was running in parallel and used the remaining heap (If yes you may find the other process number in dev_w9 trace file when dump ocurred). Check this point and let us know.
Regards.
Prithviraj
Hi Prithviraj,
I had changed the parameters after attached dump and re-tried.
Now, parameters are set,
abap/heap_area_total = 256000000000
abap/heap_area_dia = 2000000000
abap/heap_area_nondia = 0
abap/heap_limit = 40000000
ztta/roll_extension = 64000000000
em/initial_size_MB = 128000
And at last dump, memory was used;
Extended memory (EM) ...... 64014287042 (64GB)
Assigned memory (HEAP) ...... 155976325872 (155GB)
During the processing, no process other than small periodic jobs were running and no message about other processes in developer trace.
Regards,
Megumi
Hi Megumi,
Looking at the parameters and the dump is it pretty clear that heap is not used to the maximum value set.
Possible reasons as I said earlier; one other processes running parallel consumes the rest of the memory, second memory used earlier is not released to SAP by OS yet.
First check the history of heap utilization in ST02 if it has ever reached to maximum limit set i.e. 256 GB.
Try restarting the system and then while executing the process check ST02 (task manager at OS to see memory usage) with trace on the running process. If error faced pleease upload the ST22 dump, work process log file and trace generated.
Regards,
Prithviraj.
Hi Prithviraj,
I have restarted ERP instance, and also OS.
I checked commited memory decreased on shutting down ERP and increased again after started the job. Virtual memory of Windows was set to more than 600GB.
For max heap memory used in ST02, it was bit more to 155GB but less than 160GB.
Shortdump message happened the same line of the same program, EM and Heap memory usage figure was a bit different, and storage request size figure was totally same as I wrote.
Attached is workprocess trace at the termination and restarted.
Any clue is there?
Regards,
Megumi
Hi,
The only option I'm left with is imposing the memory ustilization for nondia work process to utilize complete heap that is allocation in the system.
Set the below parameter
abap/heap_area_nondia = 250 GB and give a try.
You can use RSMEMORY report to do this (without restarting the system).
Regards,
Prithviraj.
Hi Megumi,
Check your kernel version. SP8 of NetWeaver 7.40 requires a minimum of kernel 7.42, whereas SP7 worked with kernel 7.41. If you aren't updating your kernel to 7.42 you will have these sorts of problems. I believe you have to include the kernel update and the SP update together in the same SUM queue. See Note 2107721 (SAP NetWeaver kernel 741/742: List of supported ABAP versions) for more information.
Regards,
Matt
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi,
How to increasing extended memory (em/initial_size_MB) for time being, you then revert it to its intial value after the activity.
Also look in to the sapnotes below...
1967624 - External SAP HANA view: SAPKW74007 corrections
2047471 - UDO report for SAP Note 2043430 ("Enhancement of the InfoProvider Properties Maintenance")...
Regards,
Prithviraj.
Hi,
Many a time upgrading kernel avoids such issue.
Regards,
Prithviraj.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Megumi
1. Do the full system restart and then try again in the same phase.
2. Check this SAP Note for SAPLRSDG_IOBJ_DB_READ
2060453 - Long runtime in SAPLRSDG_IOBJ_DB_READ (MultiProvider with large number of PartProviders)
BR
SS
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks Sriram,
1. I reboot the system from OS several times but it didn't work.
2. I applied the note you proposed and retried, but it didn't work either...
Though the job ran bit longer, it stopped the same line of the program, LRSDG_IOBJ_DB_READF60 in SAPLRSDG_IOBJ_DB_READ.
If you hit any other idea, please advise.
Regards,
Megumi
User | Count |
---|---|
91 | |
10 | |
10 | |
9 | |
9 | |
7 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.