cancel
Showing results for 
Search instead for 
Did you mean: 

SAP HANA Index server crashed due to Memory issue

Former Member
0 Kudos

Hi All,

We are facing memory issues in one of the SAP HANA server.(SAP HANA Revision 47 on Linux)

The index server has crashed and when I tried to restart the entire instance, only name server and pre processer are running but the index server, statistics server and hdbxsengine are failing to start.

hdbdaemon is in initializing state.

when I checked the logs I could see memory allocation issues.

PFB the log of the index server

==============================

[IPMM_MEMORY]  IPMM information: (2013-02-15 12:57:55 000 Local)

Inter Process Memory Management

IPMM shared memory segment: SAP_IPMM_0009_H01

SId= Process secondary identifier

AB=  Allocated bytes by allocators for heap and stacks (inclusive allocator cache)

U=   Used bytes for heap and stacks (without allocator cache)

FSL= Free segment list size

PAL= Process allocation limit

TPAL=Temp process allocation limit

Blocked=Memory (TPAL!) blocked/reserved for this process

W   =Weight of process

[0] PID=36070, SId=12415259, compactors active, alive, process name: hdbnameserver

    AB=1075575428b (1gb), U=1225833963b (1.14gb), FSL=0b, PAL=487925253734b (454.41gb), TPAL=1071050752b (1021.43mb), W=100

    ! Executing shrink. Process with SlotID 3 requests 466902252b (445.27mb) to shrink, 4524676b (4.31mb) to reserve and garbage collection.

[1] PID=36199, SId=12415760, compactors active, alive, process name: hdbpreprocessor

    AB=122073088b (116.41mb), U=273695759b (261.1mb), FSL=0b, PAL=487925253734b (454.41gb), TPAL=122073088b (116.41mb), W=100

[2] PID=36216, SId=12415960, compactors active, alive, process name: hdbindexserver

    AB=931041280b (887.91mb), U=1102010388b (1.2gb), FSL=0b, PAL=487925253734b (454.41gb), TPAL=931041280b (887.91mb), W=100

[3] PID=36224, SId=12416061, compactors active, alive, process name: hdbstatisticsse

    AB=135325052b (129.5mb), U=304525489b (290.41mb), FSL=0b, PAL=3087007744b (2.87gb), TPAL=135325052b (129.5mb), W=100

    !! Requesting shrink...

[4] PID=36243, SId=12416162, compactors inactive, alive, process name: hdbxsengine

    AB=41893888b (39.95mb), U=?, FSL=?, PAL=3087007744b (2.87gb), TPAL=41893888b (39.95mb), W=100

Global max allocation limit=3221225472b (3gb), cached sum of allocation limits=3087007744b (2.87gb)

#checks for cached allocation limit sum=721, #terminated processes since last reset=0

#processes=5, sum of temp process allocation limits=2301384060b (2.14gb), cached sum=3087007744b (2.87gb), weight sum=500, cached weight sum=500

cached shared memory usage=785623684b (749.22mb), temp shared memory allocation limit=785623684b (749.22mb)

IPMM reservation=0b, emergency reservation=134217728b (128mb)

3 early exits since creation of IPMM, provide-memory-counter=85

Provide memory 83 failed (PID=36216), only 3863932b were free.

Users respected in shm calculation: [1002]

[OK]

Please advice what needs to be done, because I could see lot of free memory available at OS level but some how the application is not able to allocate free memory to the server processes.

Thanks in advance

Regards,

sampath

Accepted Solutions (0)

Answers (4)

Answers (4)

0 Kudos

HI same issue on my side please advice

Former Member
0 Kudos

We are facing the same issue. We have REV48 on AWS and index server crashes on restart.

After analyzing Crash logs, looks like Max Allocation Limit and cached sum of allocation limits are very close.

How can we resolve this. We may be fine to drop some of the old records but even to do that need to have system running once.

Any Suggestions

sourabhchordiya
Explorer
0 Kudos

Was there any resolution from SAP on this ? Facing same issue in SP06 rev61

lbreddemann
Active Contributor
0 Kudos

Be aware that out out memory issues are not due to a single specific bug or SAP HANA revision.

OOO situations are a generic error situation that can have many different causes.

Therefore the problematic situation needs to be analysed individually for each case.

So, if you find that this is happening on your system and don't understand why, please have a support incident opened for this.

- Lars    

lbreddemann
Active Contributor
0 Kudos

Please open a support message for these kinds of requests.

- Lars