[Beowulf] Power Cycling Question

[Rémy Dernat]

Hi,

I just saw this thread right now. I do not have the time tight now to 
read the whole thread, so, sorry if I am missing / describing something 
already discussed.

I also created a plain bash code to handle power saving on our cluster 
many years ago on SGE.

Le 17/07/2021 à 02:44, Skylar Thompson a écrit :
> We had tried engineering our custom "green cluster" automation with Grid
> Engine years ago where we would shutdown idle nodes until they were needed,
> but doing it independently of the resource manager was far too complicated
> for us to maintain, especially since it was all cost and no benefit for us
> with our power and cooling charges being absorbed through a flat overhead
> rate.

However after a big code update on SVN (before this git repo), the code 
did not work as expected and I decided to let that code as it was. It 
was also too hard to maintain for me, even if it is not really very 
complicated.

The main idea is to setup an availability level for a cluster. Let's say 
you may want 100% of active nodes + X nodes doing nothing that are ready 
to receive jobs + Y nodes poweroff (nowadays I would choose hibernate 
instead). That is the availabity level. You may decide what nodes should 
be started first, how many nodes you need to boot at a time, what queue 
to exclude...

I think all this stuff is already present SLURM, and pretty well integrated.


About suspend/hibernate, Ubuntu does this by default from Focal with 
systemd Units sleep.target, suspend.target, hibernate.target, 
hybrid-sleep.target (if you have a desktop env, even Gnome has some 
hibernate functions (see org.gnome.settings-daemon.plugins.power...).

Best regards



-- 
Rémy Dernat
Chef de projet SI
IR CNRS - ISI / ISEM
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_0xCF65C2676B3A1402.asc
Type: application/pgp-keys
Size: 3070 bytes
Desc: OpenPGP public key
URL: <http://beowulf.org/pipermail/beowulf/attachments/20210818/50524179/attachment-0001.key>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature
Type: application/pgp-signature
Size: 840 bytes
Desc: OpenPGP digital signature
URL: <http://beowulf.org/pipermail/beowulf/attachments/20210818/50524179/attachment-0001.sig>