[Linux-HA] Ethernet link down NO failover;(

[Andre Heine]

Hi all,

Am Freitag, 7. September 2007 schrieb Dejan Muhamedagic:

[...]

> > When I bring all network interfaces down, the slave take the
> > resources. All things works fine
>
> There is a difference between having cables removed (failover
> doesn't work) and bringing interfaces down (failover works)?

When all interfaces are disconnected (rcnetwork stop/restart)
the slave get the resources...

The difference is, that ifconfig show me the interfaces (bond0, 
vlan, etc) when I remove the cables.

But the effect is the same...

> > bond1 and  bond2 are "direct links/crossover" and I can't see
> > any production IP via this two interfaces.
> >
> > On bond0 there are two vlan interfaces. Over these interfaces I
> > can see the production network (see ping_group).
> >
> > When I removed the link for bond0 (vlan 425/725) the master
> > can't ping any ip from the ping group.
> >
> > On the master I get only  these log entries:
> >
> > -----------
> > prod_lan:prod_lan dead.
> > crmd[18174]: 2007/09/06_16:43:18 notice:
> > crmd_ha_status_callback: Status update: Node prod_lan now has
> > status [dead]
> > heartbeat[18132]: 2007/09/06_16:43:18 info: Link
> > nfs00002:vlan725 dead.
> > heartbeat[18132]: 2007/09/06_16:43:18 info: Link
> > nfs00002:vlan425 dead.
> > crmd[18174]: 2007/09/06_16:43:18 WARN: get_uuid: Could not
> > calculate UUID for prod_lan
>
> These log messages look normal given the circumstances.
>
> > Something is wrong in my config...
> >
> > Any hints?
>
> You should check that pingd attribute in the node's status
> section. Otherwise, please attach full logs.

Ok, I will do it next week. Sorry I'm out of the office:)

Thanks for the hints!!!

Best regards


Andre