Closed zepumph closed 1 year ago
I ran sudo mount -o size=10G,rw,nodev,nosuid -t tmpfs tmpfs /tmp
which accomplished the goal of exanding /tmp to 10G. This deleted any data that was in place in /tmp. I do not believe that this will persist a reboot. I poked around and saw that phet-server2 does not have a dedicated fs for /tmp. This could be problematic if something decides to save a bunch of temporary files, but it doesn't seem like there are any problems at the moment. phet-server however, does have a tmpfs dedicated to /tmp (at a whopping 126G!). But, this tmpfs is not saved in /etc/fstab so I'm stumped how it is getting remounted after reboots.
I tested this out with a reboot. As expected, the tmpfs
disappeared. @zepumph - do want a separated fs for /tmp
permanently? Or would you prefer to just keep that on the root LVM (i.e. /
in df -h
)?
@zepumph and I added a permanent fs for /tmp
in /etc/fstab
:
tmpfs /tmp tmpfs rw,mode=1777,size=12g
This mounted without a hitch using mount -a
. This seems to be working nicely, and should persist a reboot.
From https://github.com/phetsims/special-ops/issues/234 It is currently only a few MB. I think that 10GB is a good starting point.
We are encountering build failures because of this:
Over in https://github.com/phetsims/aqua/issues/145 we decided to also move the temp dir for all CT browser clients, and so we could do that for the build puppeteer instances too as a second option if thing are still failing after increasing
/tmp
to a "normal" size.@mattpen please expand
/tmp
on sparky to 10GB.