HA enqueue ID changes when ASCS restarts, why??

[S0001584827]

Hello HA Gurus...

We are setting a sandbox system in HA mode to demonstrate that the

users can preserve their transactions context even when we have to

restart the ASCS instance, either on the same Host or on the second host.

On HostA we have installed ERP2006 SR3:

ASCS10

DB Instance

DVEBMGS00

On HostB we have installed:

ERS11

D01

Before doing interuption, the ensmon shows that the ERS is

properly connected to the SCS:

Try to connect to host vsrs9scs service sapdp10

get replinfo request executed successfully

Replication is enabled in server, repl. server is connected

Replication is active

When we perform a restart of the ASCS10 instance, even if the user sessions are kept,

the transactions are reset. On the ERS11 work directory, the

file dev_enrepsrv indicates the following error message when ASCS10 restarts:

Reconnect: repl. table layout has changed, delete old one

And in fact, the number of entries in the replicated enqueue servers

are reset to 0 and the transactions are reset.

On ERS hosts:

/usr/sap/RS9/SYS/profile/dev_eq_trc_856194

-

-

---

trc file: "dev_eq_trc_856194", trc level: 1, release: "700"

-

-

---

Fri Feb 5 10:21:12 2010

WARNING: EnqId_EN_ObjShMem_Check: incoming EnqId checked NON-

identical

...

WARNING: ObjShMem_CheckAuthoritativeEnqId: expecting RESET_ALL_TAS

-SHMEM--

-

---

....

OLD: ReqOrd.TimeInSecs = 1265383188

OLD: ReqOrd.ReqNumberThisSec = 20

NEW: ReqOrd.TimeInSecs = 1265383272

NEW: ReqOrd.ReqNumberThisSec = 2

LOG GZZ=> EnqId change, transaction resets expected !

LOG GZZ=> OLD: EnqId.Time = 1265314739

LOG GZZ=> OLD: EnqId.Rand = 503970

LOG GZZ=> NEW: EnqId.Time = 1265383239

LOG GZZ=> NEW: EnqId.Rand = 467156

WARNING: EnqId_EN_ResetHandler: Call_ResetAllTransactions

Any clue ???

Stephane

Edited by: Stéphane Boutin on Feb 12, 2010 10:26 AM

[Former Member · Accepted solution]

How do you failover the ASCS to node B? Which HA solution do you use?

Cheers Michael

[S0001584827 · Accepted solution]

In fact, we would like to demonstrate that it's work by interrupting ASCS manually

before setting the automatic procedures.

[Former Member · Accepted solution]

Well i checked on a system with a standalone ASCS instance (but without ERS). The enqueues were also lost after a restart of the ASCS, so i am not sure wether this is supposed to work at all... maybe anybody else knows more.

Edit: But i can assure you, that the switchover with an ERS really works

[S0001584827 · Accepted solution]

Finally, it works if I issue a

"cleanipc 10 remove" command to

remove the shared memory segments associate with the ASCS.

If it helps you, give me some rewards points

Stephane

[Former Member · Accepted solution]

Glad to hear that! In my system it isn't working even with cleanipc, but my config is certainly different than yours.

So good luck with your further steps, cheers Michael

[Former Member · Accepted solution]

I have the same issue.

I installed DB and SCS instance on one AIX host and the ERS on another AIX host. The ERS connects to the SCS and when I issue the command

enqt pf=<ERS_Instance_profile> 11 30

the test entries are visible on ERS and SCS

But when I shutdown and restart the SCS all entries are reset. cleanipc on SCS does not change this.

Any idea what could be wrong? Any help is highly appreciated.

Best regards

Thomas