way for failover zpool (no HAST needed)

[Mikolaj Golub]

On Wed, 27 Apr 2011 14:05:11 +0200 Denny Schierz wrote:

 DS> hi,

 DS> Am Dienstag, den 29.03.2011, 23:36 +0300 schrieb Mikolaj Golub:
 >> 
 >> 2) There are complaints from watchdog. 

 DS> what happens, if the watchdog isn't available and one or both nodes are
 DS> rebooting or something else?

Without receiving complaints secondary wont switch to primary. This is done
intentionally, so the node does not make decision on its own. But you can have
several watchdogs if this really worries you.

 DS> the other thing what could happen: the connection between the host and
 DS> the SAS switch is death.

 DS> carp, ifstate and hastmon looking for the reachable IP, but not, if the
 DS> local storage is available. So I have a closer look to devd and zfs and

hastmon isn't just looking if the IP reachable. watchdog connects to a cluster
node and ask its status. So the result depends on how smart the script one use
to get the status.

 DS> shutdown in case of problems the carp interface / or whole machine, to
 DS> force a switch.

 DS> cu denny

-- 
Mikolaj Golub