[OpenAFS] Serious trouble, mounting /afs, ptserver, database rebuilding

Stephan Wonczak a0033@rrz.uni-koeln.de
Tue, 29 Jul 2008 10:56:43 +0200 (CEST)


   Hi Brandon!

On Tue, 29 Jul 2008, Brandon S. Allbery KF8NH wrote:

>
> On 2008 Jul 29, at 3:44, Stephan Wonczak wrote:
>
>> While more redundancy (i.e. a third database server) is always a good idea, 
>> it is not strictly necessary, much less 'a bad idea to run with two 
>> database servers'.
>> Christof probably was thinking about the 'split brain' problem, which does 
>> not come into play with the AFS architecture; we are proof against that 
>> one. I made a posting about this a while ago; it should be in the archives.
>
>
> Actually he's thinking about a screw condition that used to happen with 
> voting for a sync site if you have 2 database servers and the 
> lower-IP-numbered one goes missing.  I *think* it has been fixed now.

   Not really a bug. If the lowest numbered DB server goes AWOL the 
remaining server has no chance to become master. I cannot distinguish 
(without human help) wether there is a simple communication problem or if 
the lower server really is down. So it stays in RO-Mode, as it should.
   As an admin you then have the choice to go back to a single-server 
scenario, or bring the other one back online. I do agree however that this 
is not too nice, and you are probably better off with three DB servers.

 	Dipl. Chem. Dr. Stephan Wonczak

         Zentrum fuer Angewandte Informatik (ZAIK)
         Regionales Rechenzentrum der Universitaet zu Koeln (RRZK)
         Universitaet zu Koeln, Robert-Koch-Strasse 10, 50931 Koeln
         Tel: ++49/(0)221/478-5577, Fax: ++49/(0)221/478-5590