[Linux-HA] Problem with apcmastersnmp (AP7920)

Zachacker, Maik zachacker at ibh.de
Fri Aug 4 06:02:09 MDT 2006


I'm using a 2-node cluster with RHEL4 and Linux HA 2.0.6-1 with an APC
AP7920 as stonith-device.

The Cluster is configured to use crm and offers the services apache,
samba and mysql. This works fine, but fencing makes trouble.

The Nodes are connected to Port two and three of the apc.

When I stop heartbeat on one of the machines the other tries to stonith
this machine. This works, but a error message appears and the apc
switches the power permanently on and off.

/var/log/messages:
Aug  4 12:54:39 cluserver1 crmd: [17847]: info:
do_state_transition:fsa.c cluserver1: State transition S_POLICY_ENGINE
-> S_TRANSITION_ENGINE [ input=I_PE_SUCCESS cause=C_IPC_MESSAGE
origin=route_message ]
Aug  4 12:54:39 cluserver1 tengine: [4274]: info: unpack_graph:unpack.c
Unpacked transition 21: 10 actions in 10 synapses
Aug  4 12:54:39 cluserver1 tengine: [4274]: info:
te_fence_node:actions.c Executing reboot fencing operation (39) on
cluserver2 (timeout=2500)
Aug  4 12:54:39 cluserver1 stonithd: [17845]: info: client tengine [pid:
4274] want a STONITH operation RESET to node cluserver2.
Aug  4 12:54:39 cluserver1 stonithd: [17845]: info:
stonith_operate_locally::2289: sending fencing op (1) for cluserver2 to
device apcmastersnmp (rsc_id=DoFencing:1, pid=6374)
Aug  4 12:54:39 cluserver1 stonithd: [6374]: CRIT:
apcmastersnmp_reset_req: no active outlet for 'cluserver2'.
Aug  4 12:54:39 cluserver1 stonithd: [17845]: info: Failed to STONITH
node cluserver2 with one local device, exitcode = 4. Will try to use the
next local device.
Aug  4 12:54:41 cluserver1 stonithd: [17845]: info: Failed to STONITH
the node cluserver2: optype=1, op_result=2
Aug  4 12:54:41 cluserver1 tengine: [4274]: info:
tengine_stonith_callback:callbacks.c call=6374, optype=1,
node_name=cluserver2, result=2, node_list=,
action=39;21:a88bfed0-2320-4fbc-8e68-9b290bd10279
Aug  4 12:54:41 cluserver1 tengine: [4274]: ERROR:
tengine_stonith_callback:callbacks.c Stonith of cluserver2 failed (2)...
aborting transition.

It seems that I have configured something wrong.
Can anyone please give me a introduction what has to be configured and
how to get this work?

Regards,
Maik

--
Maik Zachacker
IBH Prof. Dr. Horn GmbH, Dresden, Germany 


More information about the Linux-HA mailing list