[Linux-HA] Stonith question

[Alan Robertson]

hm at seneca.muc.de wrote:
> hm at seneca.muc.de wrote:
> 
>>Alan Robertson <alanr at unix.sh> wrote:
>>
>>>Where it would be an "alas" would be if one of them was a serial link. 
>>>*That* link comes up basically instantly.
> 
> 
>>Not in this installation. The machines I ue do have serial ports (Intel
>>boards) but I need these for the terminal server. So - only 2 eth links. 
> 
> 
>>I'll give the initdead a try and report. 
> 
> 
> Hmmm - this was the typical "Vorführeffekt" as we say in Germany. I have no
> idea how that translates into English so ... 
> 
> I tried some failovers this morning with an unmodified ha.cf. Since the
> clocks were slightly less that a minute apart I synced them with NTP, and
> restarted heartbeat. The described effect did not occur again, so - the
> secondary became standby again after the primary re-joined the cluster.
> initdead=60, deadtime=30. 
> 
> Strange. Can the out-of-sync clocks cause this problem? 

It's been a very long time since heartbeat cared about time-of-day clocks. 
  And, it never cared that they were in sync.  Back in those days, they 
only cared if they jumped around.

Nowadays about all they're used for is formatting messages.  They do appear 
in heartbeat packets, but clock jumps no longer matter.


-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me claim 
from you at all times your undisguised opinions." - William Wilberforce