/tmp/backup_tmp_meaninglesslettersandnumbers

Hello,

TLDR :

  • [noob] this is my first opensuse usage day;
  • my disk was filled 100% (/tmp) by some mysterious /tmp/backup_tmp_adzpoidza directories.
  • any idea on what could have cause this ?

Longer version :

I had backup_tmp_meaninglesslettersandnumbers directories in /tmp that filled up my disk space entirely and resulted in angry users that couldn’t access their mailboxes.

Those backup directories disappeared by themselves this morning while sysadmin was investigating the problem, so everything returned to normal again (except for some missing mails since Saturday). I have to add that those directories were there for at least 24 hours.

I want to prevent this issue to happen again, so I am asking if anybody knows what could have caused this ? maybe a backup software ?

Thank you for your help.

chaouche wrote:
> Hello,
>
> TLDR :
>
> - [noob] this is my first opensuse usage day;
> - my disk was filled 100% (/tmp) by some mysterious
> /tmp/backup_tmp_adzpoidza directories.
> - any idea on what could have cause this ?
>
>
>
> Longer version :
>
> I had backup_tmp_meaninglesslettersandnumbers directories in /tmp that
> filled up my disk space entirely and resulted in angry users that
> couldn’t access their mailboxes.
>
> Those backup directories disappeared by themselves this morning while
> sysadmin was investigating the problem, so everything returned to normal
> again (except for some missing mails since Saturday). I have to add that
> those directories were there for at least 24 hours.
>
> I want to prevent this issue to happen again, so I am asking if anybody
> knows what could have caused this ? maybe a backup software ?

ok, hmmmm…first: backup directories do not disappear by themselves
while sysadmin are investigating a problem–so, you should very much
question the honesty of any sysadmin who tells such a tale…

my guess is the best way to prevent this issue from happening again
is to hire a competent, experienced linux/unix sysadmin who knows how
to properly setup redundant, assured backup systems and monitoring
mechanisms to alert him/her of impending crisis (like a full file
system)…that is not so very difficult to do–i mean it is easily
possible to set the system to send an email or SMS notifying that the
system is (say) 70% full and climbing…

there are a number of recognized training programs leading to linux
system administrator certification…and, please note that the
typical Microsoft Certified Systems Engineer is totally unprepared to
administer linux servers…

by the way, i guess your system is running SUSE Linux Enterprise
Server, if so you should probably ask your question over in the NOvell
forum, here:
http://forums.novell.com/novell-product-support-forums/suse-linux-enterprise-server-sles/


DenverD
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]

Thank you DenverD.

We don’t run Suse Linux, we run OpenSuse 11.3, and the mail server is a Virtual OpenSuse 11.3 server (we run xen).

> ok, hmmmm…first: backup directories do not disappear by themselves
while sysadmin are investigating a problem–so, you should very much
question the honesty of any sysadmin who tells such a tale…

Yeah, that was a guess too. I guessed they disappeared by themselves but what happens generally (I guess, again :slight_smile: ) is that applications that need to create temporary data do so in /tmp. So I guess there was some ongoing application that created those directories and then cleaned them up on quit. Does this sound plausible ?

> my guess is the best way to prevent this issue from happening again
is to hire a competent,

Yes of course, the sysadmin left us a few days ago and now we’re on our own. We will try our best until the next sysadmin is hired.

chaouche wrote:
> that applications
> that need to create temporary data do so in /tmp. So I guess there was
> some ongoing application that created those directories and then cleaned
> them up on quit. Does this sound plausible ?

possible i guess, but i don’t know of a backup program that would run
for “at least 24 hours” and then by sheer coincidence just happened to
quit and tidy up after itself while the sysadmin was looking for the
full disk problem…

nah, that sequence of events is not plausible to me…


DenverD
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]

Will scan through bash_history to see if there’s anything interresting