SCIENTIFIC-LINUX-USERS Archives

July 2011

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Thomas Hartmann <[log in to unmask]>
Reply To:
Thomas Hartmann <[log in to unmask]>
Date:
Tue, 26 Jul 2011 14:26:56 +0200
Content-Type:
text/plain
Parts/Attachments:
text/plain (51 lines)
Hi,

are there any standard daemons (time-based or system-triggered) that
purge /dev/shm ?
I noticed that the complete tmpfs on /dev/shm is cleaned at some points,
which is a bit unlucky when some processes have files located there. I
monitored the pruges but found no obvious pattern.
For the monitored machine the purges take place during early morning but
not at fixed times (and all checked crons seem to be innocent).
As for the memory consumption (as my second suspected trigger) I am not
sure if it is significant

For example, my /dev/shm was purged today morning between 3:00-3:05 and
the memory usage around that time looked as:

Tue Jul 26 02:45:01 CEST 2011
MemFree:       2113876 kB
Active:       14873624 kB
Inactive:      6345556 kB
Tue Jul 26 02:50:01 CEST 2011
MemFree:       1836656 kB
Active:       15255568 kB
Inactive:      6194388 kB
Tue Jul 26 02:55:02 CEST 2011
MemFree:        628992 kB
Active:       15924712 kB
Inactive:      6729112 kB
Tue Jul 26 03:00:01 CEST 2011
MemFree:        770884 kB
Active:       15472048 kB
Inactive:      7046180 kB
Tue Jul 26 03:05:01 CEST 2011
MemFree:       1403260 kB
Active:       16327832 kB
Inactive:      5556164 kB
Tue Jul 26 03:10:01 CEST 2011
MemFree:        848168 kB
Active:       16849468 kB
Inactive:      5600408 kB

(with tmpfs up to 20000m from a total memory of 24GB)

What is the standard kernel behaviour to treat /dev/shm if the free
memory reaches 0? -- at least in my coarse time binning the free memory
seems to get depleted  (I would have expected the kernel to start
swapping onto disk...)
Or is there probably another explanation why /dev/shm is purged?

Cheers and thanks for all ideas?,
  Thomas

ATOM RSS1 RSS2