Updater selhal: Failed to fork command /tmp/updater-busybox-jGcFEp/busybox: Out of memory ×


#1

I got this automatic notification on 15 january. Anyone knows what it means and how to fix it?

(Since then, I receive downtime notifications emails telling me “Firewall: 24 hours”. I don’t know if it is related.


Firewall down? Or just the data collection?
#2

On the router shell, I get:

DIE:Failed to fork command /etc/updater/hook_preupdate/05_schnapps.sh: Out of memory

Current memory:

root@turris:~# free
             total       used       free     shared    buffers     cached
Mem:       1030732    1007420      23312     381108       7300     416300
-/+ buffers/cache:     583820     446912
Swap:            0          0          0

#3

Could you have lots of stuff in /tmp, say about 400 MB?

EDIT: it’s a bit confusing, but kernel counts tmpfs into “cached” even though it isn’t possible to drop the cache. The whole thing is made worse by free from busybox missing IMO the most useful column “available”.


#4
root@turris:/tmp# du -sh .
17.1G	.

Wow. Time to do some rm.


#5

Strangely, one of the files in /tmp is more than one year old. /tmp is not cleared at reboot?


#6

That’s really weird. Normally /tmp/ is tmpfs mounted at size of 50% RAM. I don’t know… in general I find df to be more reliable for this than du; perhaps it could be confused by something unusual like bind-mounts, but I have no idea if anything like that is used on /tmp/ in Omnias.


#7

Yes, the problem came from other filesystems mounted under /tmp.