SLD (Java based SAP system) won't restart after server move - memory issue?

[Former Member]

Hi all

We've just moved a system (Java based SLD) from one physical server to another. The VM name is identical. The database (Sybase ASE) came up with no probs, and the Java SCS, DAA and web dispatcher all seem to have started ok. All the relevant filesystems seem to be present and permissions look good. The main instance won't come up though.

The errors are seen in the sapstart.log:

(22730) **** 2017/04/21 10:22:56 Child 22812 terminated with Status 2 . **** (22812) **** 2017/04/21 10:22:56 No RestartProgram command for program 0 ****

And more detail in dev_jstart:
M [Thr 140555251079040] *** ERROR => MsINiWrite: NiBufSend (rc=NIETIMEOUT) [msxxi.c 2766] M [Thr 140555251079040] *** ERROR => MsIDetach: MsINiWrite (rc=MSENILAYER) [msxxi.c 1560] M [Thr 140555251079040] SiSelEPSet: sock 9 was already closed but still in this set; remove now

M [Thr 140555251079040] SiSelEPRemove: sock 9 was already closed and valid again F [Thr 140555251079040] *** LOG => connecting to Message Server. F [Thr 140555251079040] *** LOG => Run level 0 completed. F [Thr 140555251079040] *** LOG => Instance state is "Instance terminated" (STOPPED @ 0, INACTIVE). I [Thr 140555251079040] MPI: dynamic quotas disabled. I [Thr 140555251079040] MPI init, created: pipes=4000 buffers=2718 reserved=815 quota=10%, buffer size=65536, total size MB=170 M [Thr 140555251079040] *** WARNING => NiSelIGetMode: invalid hdl 31 [nixxsel.cpp 1578] F [Thr 140555251079040] *** ERROR => SfCEventDispatcher::remHandler: mpSelSet->getMode failed (rc = -8 NIEINVAL) [sfxxevt.hpp 3745] etc M [Thr 140555251079040] *** ERROR => MsINiWrite: NiBufSend (rc=NIETIMEOUT) [msxxi.c 2766] M [Thr 140555251079040] *** ERROR => MsIDetach: MsINiWrite (rc=MSENILAYER) [msxxi.c 1560] I [Thr 140555251079040] *** ERROR => FcaInit: already initialized [fcaxxcomm.c 956] I [Thr 140555251079040] *** ERROR => ShmCleanup(62) failed 3 [mpixx.c 4609] F F **

F *** ERROR => Cannot connect to Message Server.

My first thoughts was that it was a shared memory error - we often have this with systems on our AIX enviroment. I shut down the main instance and the SCS, ran cleanipc etc and was going to run /usr/sbin/slibclean but then remembered this was a Linux system so can't!

Anyway, several restarts later, it's still the same situation. I'm a bit rusty with Java systems... is there perhaps something else I should doing?
Could this be related to the server move (as far as I know though, everything uses the logical VM name, not the physical server name, so it really shouldn't matter)?
Oh, and jcmon gives a "Can't open shared memory administration (rc = 3)" error.
A sappfpar check shows no errors.

Thanks

Ross

[Former Member · Accepted solution]

The issue was down to a mistake in DNS after the move; the virtual server name for the Java SCS had been misspelt with an 'a' (due to us normally having an ABAP SCS) rather than a 'j' (i.e. Java) so the app server couldn't find the SCS. Wasn't at all clear from the logs though!

[Asha_Pillai]

Hello Ross,

The instance is not able to connect to message server. Please check dev_ms log from /usr/sap/<SID>/SCSXX/work and default traces from /usr/sap/<SID>/<instance>/j2ee/cluster/serverX/log. This might be related to server move.

Regards, Asha