[Linux-HA] start/stop problem

Markus Kratzer markus.kratzer at inode.info
Thu Feb 9 08:11:40 MST 2006


hi,

following config:
a 2 node active/active cluster
nodeA ipA diskA resourcesA (nfs-kernel-server / tftpd / dhcpd)
nodeB ipB diskB resourcesB (pgsql / slapd)
heartbeat version 2.0.2 (no crm, i need ipfail)
debian 3.1

the problem:
when both nodes are online and i do a /etc/init.d/heartbeat stop on
nodeB ha-log says that ALL resources will be released, also resources
of nodeA ... (that's not the problem)
after heartbeat has shut down, nodeA says it has to acquire resourcesA
again (they're running) and tries to start the nfs-kernel-server, this
ends up in a:

ResourceManager[3451]:  2006/02/09_15:02:29 ERROR: Return code 1 from
/etc/init.d/nfs-kernel-server
ResourceManager[3451]:  2006/02/09_15:02:29 CRIT: Giving up resources
due to failure of nfs-kernel-server

after this, nodeA is releasing resourcesA, and then continiues starting
resourcesB.

this behaviour occurs only when /etc/init.d/heartbeat stop is used, with
hb_takeover and hb_standby everything's working fine ...

my questions:
why is nodeA acquiring running resources again ?
what can i do to avoid this ?

thx
markus





More information about the Linux-HA mailing list