[Triolith-users] Triolith unavailable April 3rd

[Mats Kronberg]

Triolith is now partially available. The login nodes are available and
short jobs can be run.

We might need to stop Triolith again for a short time later today or
tomorrow to make some configuration changes to the shared storage system
(due to the move of the Gamma cluster).

To make this possible we will for now only allow short jobs to start on
Triolith, and you might be logged out from the login nodes at short notice.

Triolith should be back to normal later today or tomorrow.


-- 
Mats Kronberg, NSC Support <support at nsc.liu.se>


On Mon, Mar 20, 2017 at 3:38 PM, Mats Kronberg <kronberg at nsc.liu.se> wrote:

> Dear Triolith users,
>
> Triolith will be unavailable to all users from April 3rd at 09:00 CEST
> and for at least 8 hours.
>
> No access to the cluster will be possible during the downtime. All
> users will be logged out. Jobs that cannot finish before the downtime
> will not be started until after the downtime is over.
>
> During this downtime we will be performing file system maintenance and
> also reduce the size of Triolith (see below for details).
>
> It is difficult to predict how long the work will take. Due to a bug
> in the storage software, we need to run a full file system check and
> possibly a limited repair. We estimate at least 8 hours.
>
> We will inform you when Triolith is available again, and if the
> maintenance takes longer than 8 hours, keep you updated on its
> progress.
>
>
> Details on the Triolith size reduction
> --------------------------------------
>
> As many of you already know, due to reduced funding from SNIC all
> Medium and Large allocations on Triolith will be reduced by 40% of
> their current size from April 1st, 2017.
>
> We are not just reducing project's allocations, we are also reducing
> the number of compute nodes accordingly. The nodes no longer needed
> will be removed from Triolith and re-used in another system.
>
> At this time, we believe that Triolith will continue to run at this
> reduced size until a replacement system is available, or at least
> until July 1st, 2018.
>
> The part of Triolith used by normal SNIC projects will be reduced from
> 1536 nodes to 960.
>
> The number of fat nodes will likely be reduced from 56 to 46.
>
> The number of GPU nodes will likely be reduced by one (to 4).
>
> The number of huge(2), DCS(12), WLCG(52), devel(8) and phi(3) nodes
> will remain unchanged.
>
>
> --
> NSC Support <support at nsc.liu.se>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.nsc.liu.se/mailman/public/triolith-users/attachments/20170403/5260c2ea/attachment.html>