HTCondor Project List Archives



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

Re: [Condor-devel] Condor High availability: CREDD setup



Hi Alexandre

I am actually not familiar with CREDD deamon and do not know if this service
semantics will work correctly with the primary-backup approach of the Condor
HA solution. I also don't know how its state is maintained (stateless, files
on NFS, ...), which makes a big difference for HA.

You would need to ask Condor experts on this.

Sorry I could not help more,

--- Gabi Kliot ---


Quoting Alexandre Fayolle <alexandre.fayolle@xxxxxxxxxx>:

> Shalom Gabi,
> 
> I guess that you are or were involved in Condor's HA implementation, so maybe
> 
> you know the answer to my question below: where can I find information about
> 
> the setup of Condor CREDD in a HA context ?
> 
> Thanks for your time,
> 
> Best regards
> 
> Alexandre Fayolle
> 
> On Tuesday 22 March 2011 04:58:57 Gabi Kliot wrote:
> > It is actually nice to hear that five years later, the HA technology is
> > being actively used by Condor users. Tells something about its quality and
> > importance!
> > 
> > --- Gabi Kliot ---
> > 
> > Quoting Alexandre Fayolle <alexandre.fayolle@xxxxxxxxxx>:
> > > Dear list,
> > > 
> > > I asked the question yesterday on condor-users[1] and further research
> > > showed
> > > 
> > > that a similar question asked in July[2] last year received no answer on
> > > that
> > > 
> > > list, so I'm turning to condor-devel in hope someone here can have an
> > > answer
> > > 
> > > on this: is there support for the High Availability of the CREDD daemon
> > > in Condor? There is no mention of CREDD in the manual section pertaining
> > > to HA.
> > > 
> > > 
> > > I'm using 7.4, but could upgrade to 7.5 if necessary.
> > > 
> > > >From reading the source code, I suspect that the answer is 'there is no
> > > 
> > > support' (the way the CREDD_HOST variable is used indicates that it does
> > > not
> > > 
> > > support being set to a list of hosts, which would be a first step
> towards
> > > handling failover (in the same vein as the COLLECTOR_HOST variable).
> > > 
> > > Are there any recommendation for handling that daemon in the context of
> > > HA?
> > > 
> > > Thanks in advance for any hint or advice on that matter.
> > > 
> > > [1]
> > > https://lists.cs.wisc.edu/archive/condor-users/2011-March/msg00057.shtml
> > > [2]
> > > https://lists.cs.wisc.edu/archive/condor-users/2010-July/msg00183.shtml
> > 
> > _______________________________________________
> > Condor-devel mailing list
> > Condor-devel@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/condor-devel
> 
> -- 
> Alexandre Fayolle                              LOGILAB, Paris (France)
> Formations Python, CubicWeb, Debian :  http://www.logilab.fr/formations
> Dיveloppement logiciel sur mesure :      http://www.logilab.fr/services
> Informatique scientifique:               http://www.logilab.fr/science
>