SAP HA ASCS Instance service Shared Memory Problem

[Former Member]

Hi Gurus,

I have completed the setup of ASCS and ERS instance for SAP HA. The system was working fine, but yesterday when I was trying to start the ASCS instance, it shows following error in the dev_enqsrv log file.

ShadowTable:attach: ShmCreate - pool doesn't exist

Enqueue: EnqMemStartupAction Utc=1267507616

Enqueue Info: replication enabled

Enqueue Info: enque/replication_dll not set

ShadowTable:attach: ShmCreate - pool doesn't exist

• ERROR => EnqRepAttachOldTable: failed to get information on old replication table: rc=-1 [enxxmrdt.h 393]

EnqRepRestoreFromReplica: failed to attach to old replication table: rc=-1

enque/backup_file disabled in enserver environment

***LOG GEZ=> Server start [encllog.cpp 493]

For resolving this error, I have performed following steps:

1. cleanipc 04 remove ( where 04 is the instance no for ASCS instance)

2. saposcol -k to stop the saposcollector and release the shared memory.

3. Also performed some steps like saposcol -d, kill , leave, quit.

After doing this steps, the same error is been generated in the dev_enqsrv log file. The only problem with this is the ASCS instance is starting , but while checking the log file available.log, it is changing the staus from available to unavailable every 30 secs.

Appreciate any response.

Regards,

Raj.

[nelis · Accepted solution]

What Operating System is this and did you apply the SAP recommended changes from Note 171356 - SAP software on Linux: Essential information for your specific OS ? Usually there are kernel changes made that configure shared memory.

Nelis

[Former Member]

Hi,

important to know is your NW kernel version and the setup of your file systems.

The error message could be no real error. It could be just the indication that there is no replicated enqueue table available and therefore the enqueue server of the starting ASCS instance is not able to attach to any existing table. It would create a new one. But: If you start your ASCS on the node where the ERS was running and you see that error message, that indicates that the replicated enqueue table couldn't be attached on your starting enqueue server, which is a critical error in a failover scenario.

Check following help.sap.com site for details how enqueue table replication works:

http://help.sap.com/saphelp_nw70ehp1/helpdata/en/47/e023f3bf423c83e10000000a42189c/frameset.htm

The available.log has this confusing entries if you have your /usr/sap/<SID>/<Instance>/work directory on a clustered file system which is present on both nodes. In that case, the sapstartsrv of the ASCS on the active node will write to the log file that the instance is available, whereas the sapstartsrv of the ASCS on the passive node will write to the log file that the instance is not available. It's a bug, but there are several possibilities to workaround.

If you need documentation of SAP HA on SLES or RHEL, I can post the links of the whitepapers here.

Best regards,

Matthias