cmcheckconf coredumps when checking nodes

[Danny Petterson - DK]

Hi Gurus!

Got two identical servers, with SG 11.19, 11.31.0909. On both I have identical conf-files. However, the nodes behave differently when I check the configuration:

Node2 works fine.
Node 1 core-dumps with this:
cmcheckconf -k -v -C /etc/cmcluster/myconf.conf
Begin cluster verification...
Checking cluster file: /etc/cmcluster/orabiz.conf
Defaulting MAX_CONFIGURED_PACKAGES to 300.
Checking nodes ... Abort(coredump)

Its the same version of the cmd on both nodes, Ive checked with md5sum, what and string. Everything looks identical. Same patches etc.

I COULD create the cluster on Node2, but I don't like the behavior of this, and can't help but worrying that other parts of the clustersoftware will screw on on the node with the problem.

Any clues?

Thanks in advance.

Greetings from
Danny Petterson

[Dennis Handly]

What does "file core" show?
Can you get a stack trace from gdb?

[Danny Petterson - DK]

Hi Thanks for your reply.

gdb:
Core was generated by `cmcheckconf'.
Program terminated with signal 6, Aborted.

#0 0x60000000c042f370:0 in kill+0x30 () from /usr/lib/hpux32/libc.so.1


file:
core: ELF-32 core file - IA64 from 'cmcheckconf' - received SIGABRT

Greetings from
Danny Petterson

[Stephen Doud]

You may have stumbled across a problem that has been fixed.
Install the latest patch to A.11.19 (PHSS_40784) and also the libc cumulative patch (PHCO_39526)

If these don't resolve the condition, suggest you open a case with the Support Center for deeper investigation.

[Danny Petterson - DK]

Hi Gurus!

Thanks for the input - I implemented the patches, but still no go - the exact same error - even tried some Microsoft Support, and restarted the server after patching - still no-go.

I will open a case with HP Support, and let you know what happens....

Again, thanks.

Greetings
Danny

[Dennis Handly]

>gdb: Core was generated by cmcheckconf.
Program terminated with signal 6, Aborted.
#0 0x60000000c042f370:0 in kill+0x30

You need to provide all/more frames. All signal 6 show kill(2).

Hopefully if you get SIGABRT, there should be a message saying why.

[Kapil Jha]

Hi Danny,

Are you able to run the package on this machine.
If no then what is the error in package log file.

BR,
Kapil+

[Danny Petterson - DK]

Thans for the replys

First:
Core was generated by `cmcheckconf'.
Program terminated with signal 6, Aborted.

#0 0x60000000c042f370:0 in kill+0x30 () from /usr/lib/hpux32/libc.so.1
(gdb) bt full
#0 0x60000000c042f370:0 in kill+0x30 () from /usr/lib/hpux32/libc.so.1
No symbol table info available.
#1 0x60000000c02cc5e0:0 in raise+0x120 () from /usr/lib/hpux32/libc.so.1
No symbol table info available.
#2 0x60000000c03eccd0:0 in abort+0x170 () from /usr/lib/hpux32/libc.so.1
No symbol table info available.
#3 0x60000000c25f5320:0 in cl_cassfail+0x210 ()
from /usr/lib/hpux32/libsgcl.so
No symbol table info available.
#4 0x60000000c243ac90:0 in udp_search_send+0x1a0 ()
from /usr/lib/hpux32/libsgcl.so
No symbol table info available.
#5 0x60000000c2440590:0 in search+0xeb0 () from /usr/lib/hpux32/libsgcl.so
No symbol table info available.
#6 0x60000000c2445800:0 in cl_com_search+0x17a0 ()
from /usr/lib/hpux32/libsgcl.so
No symbol table info available.
#7 0x40053c0:0 in main () at cmd/cmd_config_check.c:329
No locals.


Second:
I will not create any packages on the system, until Im sure that everything is fine on both nodes.

Greetings
Danny

[Dennis Handly]

>#3 0x60000000c25f5320:0 cl_cassfail+0x210 /usr/lib/hpux32/libsgcl.so

google finds a bunch of patches for this but they all seem older than PHSS_40784:
PHSS_40150 PHSS_40147 PHSS_36998

[Danny Petterson - DK]

Argh - I have created to APA on each node - one of them had cleared the broadcast-address, which caused the coredump - thanks for you help everybody.

Greetings
Danny

[Danny Petterson - DK]

Closed....