Rob Morin wrote:
I have not seen my original email get to the list yet... but after
looking through the logs i see this on each node...
see below for log excerts...
My test involved bringing down eth0 only(heartbeat & replication),
should i have also brought down eth1 the public side of Joe(primary)
my conf file is...
logfacility daemon # This is deprecated
keepalive 2 # Interval between heartbeat (HB) packets.
deadtime 60 # How quickly HB determines a dead node.
warntime 5 # Time HB will issue a late HB.
initdead 120 # Time delay needed by HB to report a dead
node.
udpport 694 # UDP port HB uses to communicate between
nodes.
#ping 192.168.5.1 # Ping VMware Server host to simulate
network resource.
bcast eth0
You only use one connection for heartbeat communication. That is a
configuration error.
As you unplugged that interface for testing, you forced a splitbrain
situation. Read http://www.linux-ha.org/SplitBrain
Dual split brain so to speak. Your drbd replication is also done over
this link. So not only does heartbeat loose connection, but also does
drbd. In a standard setup, a not connected secondary drbd device can be
promoted disregarding the peer's drbd state.
You might want to read about dopd, too:
http://www.drbd.org/users-guide/s-heartbeat-dopd.html
It can prevent drbd splitbrain, but you need to have >1 network
connection anyways.
#baud 115200
#serial /dev/ttyS0 # Which interface to use for HB packets.
coredumps true
auto_failback on # Auto promotion of primary node upon
return to cluster.
Your comment answers your later question on what will happen when a
rebooted (stonith'd) node rejoins the cluster.
Regards
Dominik
node joe # Node name must be same as uname -n.
node stewie # Node name must be same as uname -n.
###
###
respawn hacluster /usr/lib/heartbeat/ipfail
# Specifies which programs to run at startup
# DO not use the below unless you use the /var/lib/heartbeat/crm/cib/xml
config file instead
#crm on
use_logd yes # Use system logging.
logfile /var/log/hb.log # Heartbeat logfile.
debugfile /var/log/heartbeat-debug.log # Debugging logfile.
Primary
--------
May 6 23:04:44 joe heartbeat: [4342]: WARN: node stewie: is dead
May 6 23:04:44 joe heartbeat: [4342]: WARN: No STONITH device configured.
May 6 23:04:44 joe heartbeat: [4342]: WARN: Shared disks are not
protected.
May 6 23:04:44 joe heartbeat: [4342]: info: Resources being acquired
from stewie.
May 6 23:04:44 joe heartbeat: [4342]: info: Link stewie:eth0 dead.
May 6 23:04:44 joe heartbeat: [4249]: debug: notify_world: setting
SIGCHLD Handler to SIG_DFL
May 6 23:04:44 joe mach_down[4283]: [4328]: info:
/usr/lib/heartbeat/mach_down: nice_failback: foreign resources acquired
May 6 23:04:44 joe heartbeat: [4342]: info: mach_down takeover complete.
May 6 23:04:44 joe heartbeat: [4342]: debug: StartNextRemoteRscReq():
child count 1
May 6 23:04:44 joe heartbeat: [4250]: info: Local Resource acquisition
completed.
Secondary
-----------
May 6 23:04:46 stewie heartbeat: [21820]: info: Resources being
acquired from joe.
May 6 23:04:46 stewie heartbeat: [21820]: info: Link joe:eth0 dead.
May 6 23:04:46 stewie heartbeat: [4946]: info: No local resources
[/usr/lib/heartbeat/ResourceManager listkeys stewie] to acquire.
May 6 23:04:46 stewie heartbeat: [21825]: ERROR: MSG[4] :
[info=req_our_resources()]
May 6 23:05:10 stewie mach_down[4953]: [6063]: info:
/usr/lib/heartbeat/mach_down: nice_failback: foreign resources acquired
May 6 23:05:10 stewie heartbeat: [21820]: info: mach_down takeover
complete.
May 6 23:05:10 stewie heartbeat: [21825]: ERROR: MSG[2] : [info=mach_down]
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
