[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[Linux-cluster] CS4/ question about fence reboot


CS4/ I've a question about the default action "reboot"
which has been choosen apparently when CS4 uses scripts such
as "fence_bullpap", and others such as fence_ipmilan, etc.

The healthy node "fences reboot" the failed one in two
phases : poweroff and then poweron.
It seems that the healthy node waits at first for
poweroff status, and then proceeds to poweron and
then again waits for power on status.
Once the CS4 on this node gets the "poweron" status, it
can recover the services previously active
on the failed node.

What about the case where the failed node has a failure
such as poweron is no more possible ?
In this case, I think the CS4 will not failover and will
be stalled, will it?

Why not considering only the poweroff status ? I would
have thought it was sufficient to takeover without any
risk of double access to shared disk space, am I wrong ?

Alain Moullé

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]