[conspire] tmpreaper and use of /tmp as scratch space

[Ivan Sergio Borgonovo]

On 07/07/2018 11:36 PM, Rick Moen wrote:
> I should correct some potentially misleading wording:
> 
>> Absent volatile-storage issues, files left in /tmp/ will
>> stick around indefinitely _unless_ there's a boot script to clean out
>> old files at startup, and/or a system maintenance script or other type
>    ^^^
>> of utility running to periodically remove old, not-recently-touched
>> files from it.
> 
> Further down, I also used the phrase 'unused files'.  In both cases, I
> really should have used the term 'unaccessed files'.
> 
> I tried really hard to not get lost in the weeds of Unix minutiae, but
> this is probably worth covering, just a bit:  Both tmpreaper and the
> system reboot scripts rely on the 'atime' time stamp, which you can
> think of as access time.  Absent special situations, atime is one of
> three time stamps kept concerning all files, and gets updated if, e.g.,
> you do an 'ls' of the directory containing the file, and not _just_ if
> you modify the file's contents.  Here is a brief rundown on the three
> timestamps: http://kerolasa.github.io/filetimes.html

meh... I took the habit of mounting stuff with noatime.
This setup became much more popular with SSD.

And I just noticed that tmpreaper is not installed by default.
Has this task been delegated to systemd?

Not that I really care, at my current job rebooting/turning off machines 
is not really an issue and I tend to do it reasonably often.

Still... to whom it may concern...

-- 
Ivan Sergio Borgonovo
https://www.webthatworks.it https://www.borgonovo.net