Hi,

On Fri, Jun 19, 2009 at 03:45:36PM +0200, Uwe Gro?kinsky wrote:
> y Folks
> 
> i am using 2 nodes with eth0, eth2 and serial
> 
> lbxx01 eth0
> 
> 
> my ha.cf looks so (on one maschine the 2cond is the same exlude ping and 
> ucast)
> 
> debugfile       /var/log/ha-debug
> logfile         /var/log/ha-log
> logfacility     local0
> keepalive       2
> warntime        5
> deadtime        10
> initdead        30
> deadping        20
> udpport         694
> #bcast          eth1 eth2
> ucast   eth0    10.23.5.212
> ucast   eth2    130.144.9.14
> ucast   eth0    10.23.5.212
> ucast   eth2    130.144.9.14
> #respawn hacluster /usr/lib/heartbeat/ipfail
> #apiauth ipfail gid=haclient uid=had
> ping            130.144.9.14
> ping            10.23.5.212
> baud            38400
> serial          /dev/ttyS0
> auto_failback   on
> node            lbcisxx01
> node            lbcisxx02
> #debug          1
> 
> 
> each node has a special pool configured in haresources, so the load was 
> mixed on both 2 nodes like this example (30 IP's per node)
> 
> haresources (example as short version)
> 
> lbxx01 IPaddr::130.144.9.90 ldirectord::www01
> lbxx02 IPaddr::130.144.9.80 ldirectord::www02
> 
> my Heartbeat (node1 and node2) is running well, if one node have a 
> problem so the second node will handle the rest from the pool (and 
> himselfe).
> i mean ... if on one node heartbeat are dead or machine was offline so 
> the other node will make the takeover
> so the machie will going down or heartbeat is going down or crashed
> 
> now it will go to complicate for me !!
> 
> I will configure both nodes that in case of , eth0 or eth2 are dead (but 
> ping was also possible), the second node will handle the pool.
> or
> with the serial cable connection one node have problems so also the 
> other aktive node will handle the full pool
> 
> i will check all conechtion .... if one is dead the second node should 
> handle and start the second pool on ldirectord
> 
> what is the best way to configure this ?
> is it possible on this way ?

Must say that I didn't quite get it :)  In any case, you can use
ipfail to check the connectivity, i.e. communication paths which
are important for your services, and it will initiate a failover
in case one node has better connectivity.

Thanks,

Dejan

> Thanks  (great project :) )
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to