[SOLVED] can /tmp/SBo content safely be deleted

Isix · 12-14-2010, 01:21 AM

Hi all,

After installing numerous stuff on my Slackware system, I notice I am running out of hard-drive space. I see that /tmp/SBo has about 1G of staff that I recently installed --- may I safely delete this staff?

Cheers,
IDD

rfernandez · 12-14-2010, 01:44 AM

Sure you can. Maybe you'd like to save the .tgz packages for the future, but the /tmp/SBo folder you can remove - it's where the sources are unpacked and programs compiled and installed with DESTDIR. Take a look at some SlackBuild script and you'll see what are those.

AwesomeMachine · 12-14-2010, 01:44 AM

Yes you may! You can delete everything in /tmp if you really want to. Whatever the operating system actually requires the kernel will make again.

Isix · 12-14-2010, 01:55 AM

Thanks for the guide --- now I have got that 1G space back.

catkin · 12-14-2010, 02:17 AM

Opinions vary about the best way to clean /tmp and none of them are guaranteed to cause no breakage. I remove everything in /tmp as soon as the file system containing it is mounted; this by modifying the boot script which mounts it; works for me; YMMV.

Isix · 12-14-2010, 02:48 AM

Hi catkin,

you say, "by modifying the boot script which mounts" /tmp. Which is the boot script you are referring to (is it /boot/config ?), and how did you modify it?

Tinkster · 12-14-2010, 03:37 AM

Quote:

Originally Posted by AwesomeMachine

Yes you may! You can delete everything in /tmp if you really want to. Whatever the operating system actually requires the kernel will make again.

Ummm ... no, you can't, you shouldn't. Many apps, including X11,
have temporary files in /tmp for users CURRENT sessions. Deleting
those can make "bad things" happen. It's safe to delete OLD stuff
from /tmp, things that aren't currently being accessed.

Cheers,
Tink

catkin · 12-14-2010, 03:37 AM

On Slackware it's /etc/rc.d/rc.S. When /tmp is on the / file system, the relevant part is just after comment "# Remount the root file system in read-write mode" which is mostly an if-fi to deal with it failing. I use this line after the closing fi: ( cd /tmp && rm -rf -- * .* 2>/dev/null )

sycamorex · 12-14-2010, 03:42 AM

I delete everything in /tmp when the system shuts down by adding the following to /etc/rc.d/rc.local_shutdown:

Code:

/usr/bin/find /tmp -mindepth 1 -maxdepth 1 -print0 | xargs -0r /bin/rm -rf

catkin · 12-14-2010, 03:42 AM

Quote:

Originally Posted by Tinkster

It's safe to delete OLD stuff
from /tmp, things that aren't currently being accessed.

+1 to all of Tinkster's post. Regards the quote above, the potential gotcha is that even old files and directories may be in use. This is a topic that has been much debated and, as I posted earlier, there are no guaranteed safe solutions. Prior discussions on LQ listed here.

catkin · 12-14-2010, 03:52 AM

Quote:

Originally Posted by sycamorex

I delete everything in /tmp when the system shuts down by adding the following to /etc/rc.d/rc.local_shutdown:

Code:

/usr/bin/find /tmp -mindepth 1 -maxdepth 1 -print0 | xargs -0r /bin/rm -rf

I decided against that solution in favour of doing it during boot because rc.[06] runs rc.local_shutdown early so before shutting down many other things which _might_ depend on files+directories in /tmp including MySQL, Apache, Samba, NFS server, ssh (which does use /tmp), SASL, LDAP, D-bus, networked file systems, networking, process accounting, ACPI, quotas (which does use /tmp?) ...

sycamorex · 12-14-2010, 03:56 AM

Quote:

Originally Posted by catkin

I decided against that solution in favour of doing it during boot because rc.[06] runs rc.local_shutdown early so before shutting down many other things which _might_ depend on files+directories in /tmp including MySQL, Apache, Samba, NFS server, ssh (which does use /tmp), SASL, LDAP, D-bus, networked file systems, networking, process accounting, ACPI, quotas (which does use /tmp?) ...

Thanks. Never thought of that.

Isix · 12-14-2010, 04:42 AM

After adding

Code:

cd /tmp && rm -rf -- * .* 2>/dev/null

in /etc/rc.d/rc.S after the line

Code:

# Remount the root file system in read-write mode

and then rebooting, I now have 4G of space. Wow, I needed space like cracy --- thanks!

pwc101 · 12-14-2010, 04:57 AM

Quote:

Originally Posted by catkin

( cd /tmp && rm -rf -- * .* 2>/dev/null )[/B]

I shy away from the .* wildcard because I've always though it'd expand to '..'. I tend to use .??* which forces the wildcard to expand to at least three characters (thereby omitting '..'). It still catches most dot-files: only those whose file name is fewer than three characters long don't get deleted. I have yet to come across such a file.

catkin · 12-14-2010, 05:32 AM

Quote:

Originally Posted by Isix

After adding

Code:

cd /tmp && rm -rf -- * .* 2>/dev/null

in /etc/rc.d/rc.S after the line

Code:

# Remount the root file system in read-write mode

and then rebooting, I now have 4G of space. Wow, I needed space like cracy --- thanks!

Safer to use

Code:

( cd /tmp && rm -rf -- * .* 2>/dev/null )

in case rc.S needs to have a specific current working directory (AFAIK it doesn't but best to play safe).

And it should go after the fi after

Code:

# Remount the root file system in read-write mode

not directly after it (because the / file system has been mounted read-write then -- surprising that it worked for you!).