[Linux-cluster] groupd SEGFAULT

[Ray Van Dolson]

I'm trying to set up a 2-node cluster on RHEL 5.3 POWER (ppc64).  When
I do a service cman start after configuring my cluster.conf file (via
system-config-cluster), it appears that groupd is segfaulting.

/var/log/groupd:
  1241721765 cman: our nodeid 2 name domusB.domain.com quorum 1
  1241721765 groupd segfault log follows:

There are core files generated in /.  If I install the debuginfo's and
run via gdb:

(gdb) set follow-fork-mode child
(gdb) r
Starting program: /sbin/groupd 

Program received signal SIGSEGV, Segmentation fault.
[Switching to process 19766]
0x0ff238b0 in semctl@@GLIBC_2.2 () from /lib/libc.so.6
(gdb) bt full
#0  0x0ff238b0 in semctl@@GLIBC_2.2 () from /lib/libc.so.6
No symbol table info available.
#1  0x0fe1314c in openais_service_connect () from /usr/lib/openais/libcpg.so.2
No symbol table info available.
#2  0x0fe13d60 in cpg_initialize () from /usr/lib/openais/libcpg.so.2
No symbol table info available.
#3  0x1000c96c in setup_cpg () at cpg.c:638
  error = 1516
  fd = 1241721509
#4  0x10012930 in loop () at main.c:816
  rv = 0
  i = 268367568
  timeout = -1
  workfn = (void (*)(int)) 0x80
  deadfn = (void (*)(int)) 0xfff8fa64
#5  0x1001357c in main (argc=1, argv=0xfff8fa64) at main.c:1054
  i = 4

My cluster.conf is as follows:

  <?xml version="1.0"?>
  <cluster alias="domus" config_version="4" name="domus">
    <fence_daemon post_fail_delay="0" post_join_delay="3"/>
    <clusternodes>
      <clusternode name="domusA.domain.com" nodeid="1" votes="1">
        <fence>
          <method name="1">
            <device blade="4" name="sysibmbc1.domain.com"/>
          </method>
        </fence>
      </clusternode>
      <clusternode name="domusB.domain.com" nodeid="2" votes="1">
        <fence>
          <method name="1">
            <device blade="2" name="sysibmbc1.domain.com"/>
          </method>
        </fence>
      </clusternode>
    </clusternodes>
    <cman expected_votes="1" two_node="1"/>
    <fencedevices>
      <fencedevice agent="fence_bladecenter" ipaddr="10.49.4.192" login="user" name="sysibmbc1.domain.com" passwd="pass"/>
    </fencedevices>
    <rm>
      <failoverdomains/>
      <resources>
        <ip address="10.49.6.97" monitor_link="1"/>
      </resources>
    </rm>
  </cluster>

Have I hit a bug or maybe just configured something wrong?

Ray