[Dunder-users] Re: tornado/dunder: service stop on Tue May 27 0800 CEST

[Per Lundqvist]

The service stop on Dunder is now over. All filesystems are available and 
the batch queueing system is running again.

/Per


On Tue, 27 May 2008, Per Lundqvist wrote:

> These are now available from Dunder:
> 
>    /nobackup/rossby4
>    /nobackup/rossby5
>    /nobackup/rossby6
>    /nobackup/rossby7
>    /nobackup/rossby8
>    /nobackup/smhid1
>    /nobackup/smhid3
> 
> /Per
> 
> On Tue, 27 May 2008, Per Lundqvist wrote:
> 
> > The following filesystems are now available again from Dunder and Tornado:
> > 
> >    /nobackup/rossby1
> >    /nobackup/rossby2
> >    /nobackup/rossby3
> > 
> > /Per
> > 
> > 
> > On Tue, 27 May 2008, Per Lundqvist wrote:
> > 
> > > The stop is not over yet, but the login node on Dunder is now available.
> > > 
> > > Note that no /nobackup-filesystem is available at the moment. The 
> > > remaining filesystems will be available as soon as we are finished with 
> > > the maintenance of them.
> > > 
> > > /Per
> > > 
> > > 
> > > On Mon, 19 May 2008, Per Lundqvist wrote:
> > > 
> > > > On Tue May 27 0800 CEST, _all_ distributed filesystems on Dunder and 
> > > > Tornado (i.e. /home and /nobackup/*) will be taken offline for maintenance 
> > > > of the corresponding disk servers.
> > > > 
> > > > Since this also affects /home on both systems, everybody will have to be 
> > > > logged out while the corresponding disk server is being checked - we will 
> > > > prioritize /home though, so that you may log in as soon as possible.
> > > > 
> > > > The batch queuing system will be stopped during this - i.e. we will put a 
> > > > blocking reservation that prevents jobs to run at the time.
> > > > 
> > > > The amount of time for this is uncertain, an initial estimate is that we 
> > > > are done ~ 2000 on the same day.
> > > > 
> > > > Schedule Tue 27/5:
> > > >    0800 - stop begins, everybody will be logged out from dunder and
> > > >           tornado. The batch queuing system will be stopped.
> > > >    ...	- all distributed filesystems will be unmounted from all nodes
> > > > 	  on Dunder and Tornado
> > > >    ...	- maintenance begins on all disk servers
> > > >    ...  - filesystems will be mounted again as soon as possible (/home
> > > >           prioritized)
> > > >    ...	- when /home is available you will be notified and be able to
> > > >           log in again
> > > >    ...	- when the rest of the filesystems on each respective
> > > >           filesystem are available you will be notified and the batch
> > > >           queuing system will start again.
> > > > 
> > > > /Per
> > > > 
> > > > 
> > > 
> > > 
> > 
> > 
> 
> 

-- 
Per Lundqvist

National Supercomputer Centre
Linköping University, Sweden

http://www.nsc.liu.se