Hi Pierre,
this is interesting. 
I was thinking about reducing switchover times by having a 
standby daemon running, which - once "activated" - would 
immediately be able to serve users after the active daemon 
died. 
Can you give some more information about what you did to the
daemon, so it is a standby and can be activated upon failover/
switchover?
Cheers,
Uli

> -----Original Message-----
> From: ext Pierre Sangouard [mailto:[email protected]] 
> Sent: Sunday, June 27, 2010 10:02 AM
> To: 'Lars Wetzel'; [email protected]
> Cc: [email protected]
> Subject: Re: [Openhpi-devel] Log dump for issue #1868725 
> puttopastebin.com...
> 
> 
> I actually worked on a HA solution where 2 OpenHPI daemons - 
> 1 active, 1
> standby - would talk to the active Shelf Manager through 
> RMCP. It did not
> work properly since both daemons would act on events at the 
> same time (Hot
> swap events when a board was inserted for example), competing 
> for SEL and
> SDR repository access (via IPMI Reserve commands). What we 
> ended up doing is
> having the standby daemon waiting to be activated upon 
> failover/switchover
> commands (This is of course a very high level description).
> 
> Cheers,
> Pierre
> 
> > -----Original Message-----
> > From: Lars Wetzel [mailto:[email protected]]
> > Sent: Saturday, June 26, 2010 20:06
> > To: [email protected]
> > Cc: Pierre Sangouard; [email protected]
> > Subject: Re: [Openhpi-devel] Log dump for issue #1868725 put
> > topastebin.com...
> > 
> > Hi Pierre,
> > 
> > right, I could also see problems with having two Daemons 
> connected to one
> > ShelfManager. But I'm unsure if these problems only occur with a
> > connection
> > over SMI? Can this also happen with a connection over lan via RMCP?
> > 
> > Best Regards
> >    Lars
> > 
> > 
> > On Saturday, 26. June 2010 15:11, Pierre Sangouard wrote:
> > > > -----Original Message-----
> > > > From: Frank C. Earl [mailto:[email protected]]
> > > > Sent: Tuesday, June 22, 2010 23:03
> > > > To: Lars Wetzel
> > > > Cc: Earl, Frank; [email protected]
> > > > Subject: Re: [Openhpi-devel] Log dump for issue #1868725 put
> > > > topastebin.com...
> > > >
> > > > Lars Wetzel wrote:
> > > > > I will look a little bit deeper in both files in the 
> next days.
> > > >
> > > > Thanks!
> > > >
> > > > > Don't wonder about the questions, I don't have any 
> hardware here and
> > I
> > > >
> > > > want to
> > > >
> > > > > exclude any firmware issue as far as it is possible.
> > > >
> > > > I'm familiar with the concept, trust me.  :-D
> > > >
> > > > > Btw. you mentioned you use a SHMM-500. I think it 
> will work with
> > > >
> > > > synchronous
> > > >
> > > > > SEL?
> > > >
> > > > Okay, I'll look into that one.  Do you know if the 
> Colorado Springs
> > > > CMM's  will work in that mode as well?
> > > >
> > > > > Did you have two daemons running which are connected 
> to the same
> > > >
> > > > ShelfManager?
> > > >
> > > >
> > > > To the best of my current knowledge only one daemon is 
> up.  In an HA
> > > > configuration, the monitoring and management boards would have a
> > daemon
> > > > up on both the master and
> > > > the standby, but that'd just increase IPMI traffic, 
> wouldn't it?  We
> > > > don't care overmuch about gaps in information as long 
> as it stays up
> > and
> > > > self-heals over a realistic timeframe.
> > >
> > > It's a bad idea to have 2 daemons talking to the Shelf 
> Manager at the
> > same
> > > time. They will compete for SDR and SEL access and it won't work
> > properly.
> > > In a HA configuration, only the master daemon should talk 
> to the Shelf
> > > Manager (Been there, done that a couple of years ago).
> > >
> > > Pierre
> > >
> > >
> > >
> > > 
> --------------------------------------------------------------
> ----------
> > ---
> > >--- This SF.net email is sponsored by Sprint
> > > What will you do first with EVO, the first 4G phone?
> > > Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> > > _______________________________________________
> > > Openhpi-devel mailing list
> > > [email protected]
> > > https://lists.sourceforge.net/lists/listinfo/openhpi-devel
> > 
> > --
> > -------------------------------
> > Dipl. Wi.ing.
> > Lars Wetzel
> > Uttinger Str. 13
> > 86938 Schondorf a. Ammersee
> > 
> > Tel.: 0179-2096845
> > Mail: [email protected]
> > 
> > USt-IdNr.: DE181396006
> 
> 
> --------------------------------------------------------------
> ----------------
> This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Openhpi-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/openhpi-devel
> 

------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Openhpi-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openhpi-devel

Reply via email to