To repeat: if the build dumps core, more disk space will not fix it.
It will fix the problem of the rootfs being 108% full (which may have a whole bunch of side-effects).
What I do is:
- link
/tmp to
/var/tmp-/
- link
/root to or
/home/root/
Note I use
/var/tmp-/, not
/var/tmp/ ...
/tmp/ is supposed to be `volatile', meaning the content are deemed unimportant and susceptible to deletion at any point.
/var/tmp/ is not...
Now, the first answer to this will probably be something along the lines of:
The root filesystem should be self-contained, if /var/ or /usr/ is unavailable stuff might fail
I will save on electrons by responding preemptively:
This *used* to be good advice when different filesystems used to be different physical disks, but this hasn't been the typical use case in ages. Certainly not on the desktop. The changes of
/usr/ failing and
/ still working fine are pretty small.
/usr/ is a silly filesystem anyway, the only reason it's there is because a 256K disk was full in 1972 (see:
here and
here).
/var/ is more useful, it does a lot of writes and tends to get corrupted a bit fasted on crashed etc. but the only case where this is a problem is when a)
/var/ is completely useless and b) You need to do complex stuff with just your rootfs... Both are highly unlikely, and *if* this occurs and *if* you're competent enough to end up here, then surly you can re-create a missing
/tmp/.