cancel
Showing results for 
Search instead for 
Did you mean: 

HANA DB is not starting. Indexserver OOM issue

Former Member
0 Kudos

Hello All,

need a urgent help here. Please suggest.

We have 8 GB ram on the server.

We use HANA 1.0 rev 96.

Iam trying to start my HANA db using HDB start, but fails with below error:

StartService

Impromptu CCC initialization by 'rscpCInit'.

  See SAP note 1266393.

OK

OK

Starting instance using: /usr/sap/CSP/SYS/exe/hdb/sapcontrol -prot NI_HTTP -nr 00 -function StartWait 2700 2

16.06.2016 14:04:18

Start

OK

16.06.2016 14:04:28

StartWait

FAIL: process hdbindexserver HDB Indexserver not running

the trace files shows the below error:

Memory consumption information of last failing ProvideMemory, PM-INX=117:

IPMM short info:

GLOBAL_ALLOCATION_LIMIT (GAL) = 7176265728b (6.68gb), SHARED_MEMORY = 346841088b (330.77mb), CODE_SIZE = 1664507904b (1.54gb)

PID=33701 (hdbcompileserve), PAL=7310483456, AB=999174144, UA=0, U=171275044, FSL=0

PID=33585 (hdbnameserver), PAL=7310483456, AB=1622446080, UA=0, U=722165568, FSL=0

PID=33698 (hdbpreprocessor), PAL=7310483456, AB=929103872, UA=0, U=94172512, FSL=0

PID=33765 (hdbindexserver), PAL=7310483456, AB=1614192640, UA=0, U=702658147, FSL=0

Total allocated memory= 7176265728b (6.68gb)

Total used memory     = 3701620263b (3.44gb)

Sum AB                = 5164916736

Sum Used              = 1690271271

Heap memory fragmentation: 48% (this value may be high if defragmentation does not help solving the current memory request)

Top allocators (component, name, size). Ordered descending by inclusive_size_in_use.

Host: gotsap410

Executable: hdbindexserver

PID: 33765

Failed to allocate 262144 byte.

Can someone please suggest. How to start up this.

Regards,

Vishwanath B

Accepted Solutions (0)

Answers (4)

Answers (4)

Former Member
0 Kudos

Hi ,

Change global allocation limit to 15gb or 20gb and start. Dont worry about current ram size .

Once the db is up you can revert back the global alocation limit to default value.

[memorymanager]

global_allocation_limit = 15728640

Regards,

Raghu

Former Member
0 Kudos

i set the following in global.ini... still sam error after setting tp 20 GB as below

[memorymanager]

global_allocation_limit=2000

davidebruno
Participant
0 Kudos

the valuehas to be in MB..

for 20GB you have to se 20000

Former Member
0 Kudos

yes set to 20000 only

former_member183326
Active Contributor
0 Kudos

Can you try changing the pre load of the column store tables to false.

Then stop the full system.

Start the full system.

Let me know if this works.

Also do you know the size of your column store + row store?

Former Member
0 Kudos

DiD the same.

not sure about the " size of your column store + row store " ?

did restart, same error.

former_member183326
Active Contributor
0 Kudos

Ok, well i'm completely out of ideas. It really looks as though this has never started, its just too small. The code alone on the HANA Machine needs 1.5Gbs which only leaves you with around 5GBs to start the system?

I hope someone else can help.

davidebruno
Participant
0 Kudos

sorry but this instance was started in the past?

I ask this because 8GB of RAM is very very little..

Former Member
0 Kudos

Yes, it was started in the past,  i Agree that its low ram , but we cannot go now for extesnion of RAM now, since it requires levels of approval which iam afraid will not get now

Please suggest some alternatives if possible

Regards,

Vishwanath B

davidebruno
Participant
0 Kudos

increase the Global Allocation Limit from


/usr/sap/<SID>/SYS/global/hdb/custom/config

Former Member
0 Kudos

i have changed it to the following ( assigned the value of RAM available here ), but still same error

[memorymanager]

global_allocation_limit = 8187720.

Please suggest

Regards,

vishwanath  B

davidebruno
Participant
0 Kudos

can you stop every HANA process, clean the shared memory, and then try to restart

former_member183326
Active Contributor
0 Kudos

Try removing the allocation limit to see if it will start then.

Can you attach the indexserver trace file?

saroj_bagai
Contributor
0 Kudos

you can check shared memory segments used with ipcs -ma  and then clean it using ipcrm

Former Member
0 Kudos

did HDB stop

then checked for any running process, found none.

did HDB start. stil same error as above.

saroj_bagai
Contributor
0 Kudos

check free memory on the host using free -m and check for orphan shared memory segments and then clean them up  using -ipcrm

Former Member
0 Kudos

did ipcs -ma , found lot of shmids owned by sidadm. Cleared it using ipcrm -m , then started. Still same error.

Also cleared the global_allocation_limit as bruno suggested, still same error.

saroj_bagai
Contributor
0 Kudos

If possible try to restart the Host Machine

davidebruno
Participant
0 Kudos

can you attach the indexserver trace?

Former Member
0 Kudos

Hi,

removed global_allocation_limit. No luck.

Attached the trace file as well. Pleas let me know

Regards,

Vishwanath B

Former Member
0 Kudos

trace file attached here,

Former Member
0 Kudos

free -m output

             total       used       free     shared    buffers     cached

Mem:          7995       2813       5181        367          8       1746

-/+ buffers/cache:       1059       6936

Swap:         2053        159       1894

cleared the shared memory using ipcrm -m. Still same issue

former_member183326
Active Contributor
0 Kudos

This is unreadable?

Please try again. Please place it in the .txt format.

It seems to be encrypted.

Former Member
0 Kudos

i have attached the trace files in 3 seperate text files due to 1 MB restriction.

former_member183326
Active Contributor
0 Kudos

Can you try changing the pre load of the column store tables to false.

Then stop the full system.

Start the full system.

Let me know if this works.

Also do you know the size of your column store + row store?

davidebruno
Participant
0 Kudos

can you attach also the .ini files under

/usr/sap/<SID>/SYS/global/hdb/custom/config

former_member183326
Active Contributor
0 Kudos

Please increase the Global Allocation Limit. The Indexserver doesnt have enough the start. See SAP Note 1999997 on more information on the Global Allocation Limit.

Please be aware SAP Product Support has instant chat now available where these question can be asked in real time with a SAP HANA engineer.

Michael

former_member182967
Active Contributor
0 Kudos

Hi Vishwanath,

Only 8GB...............Why do you think you can start HANA normally?

If possible, extend your memory size.

Regards,

Ning

Former Member
0 Kudos


Hi Ning,

THis was the hardware design that was already in place for the system. It used to startup normally initially.

I assume they have configured less memeory for this server coz its a Sandbox system .

Regards,

Vishwanath B