I would like to see Phabricator move to a virtual owed cluster, or even a
bare metal cluster. However, I would not include Phabricator in a list
of "not-so-resource-consuming
services."
Phabricator definitely uses significant amount of resources. It's running
on a 12 core server with 64gb ram and 500gb raid storage, currently.
That hardware
is a little bit more than required but not by very large margin.
I've recently requested a second machine to be used for a backup
Phabricator instance, so that we can avoid long downtimes during
maintenance or hardware failure. That request is still being discussed at
https://phabricator.wikimedia.org/T131775
Phabricator is gaining improved high-availability support thanks to recent
work upstream, so it might be possible to have dual-master phabricator
nodes in the near future. See
https://secure.phabricator.com/T10751 for
upstream progress.
On Monday, April 4, 2016, Vituzzu <vituzzu.wiki(a)gmail.com> wrote:
Why not a small virtualised cluster for these
not-so-resource-consuming
services like OTRS, phab, etc?
/me runs away before writing the world-which-shouldn't be written
Vito
Il 04/04/2016 19:57, Greg Grossmeier ha scritto:
Apologies for not sending out this announcement
before hand.
Short summary: The machine that Phabricator is hosted on rebooted itself
last night due to high temperatures. It ended up just shutting itself
down.
Today we needed our DataCenter Technician to reapply the thermal paste
in an attempt to remedy the issue. That took less than 10 minutes but it
happened during the middle of the day.
Full details:
https://phabricator.wikimedia.org/T131742
And yes, we are requesting a backup machine so issues like this won't
have as much of an impact on you (our users):
https://phabricator.wikimedia.org/T131775
Best,
Greg
_______________________________________________
Wikitech-l mailing list
Wikitech-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l