Am 03/24/2013 10:36 PM, schrieb Dave Love:
Funny thing is that in procfs.c this is also done, but only as a fallback if
there
is no "Swap:" data in /proc/<pid>/smap and not as a default.
What source are you looking at? The logic is supposed to be: use
PSS+swap if available (swap from status preferably, else smaps), else
use RSS+swap, if available, else use VMsize.
I was looking at the official 8.1.3 release tar.gz
The logic I was looking at:
errno = 0;
/* Ideally, use PSS for best accuracy. */
if (pss_in_smaps()) {
snprintf(procnam, sizeof procnam, PROC_DIR "/%s/smaps", proc);
[...]
} else if (swap_in_status()) {
/* Slightly quicker if we have it -- maybe not worth bothering. */
snprintf(procnam, sizeof procnam, PROC_DIR "/%s/status", proc);
[...]
} else if (swap_in_smaps()) {
/* Last resort is to examine all the maps in smaps. */
snprintf(procnam, sizeof procnam, PROC_DIR "/%s/smaps", proc);
[...]
}
The pss_in_smaps() triggers the complete /proc/pid/smaps parsing on both RHEL5 and
RHEL6.
I think you have a pathological case, but I'll add an option, and see if
the parsing can be made faster. You should be able to turn it off
easily by rebuilding with pss_in_smaps always returning false.
I just did that (also removed swap_in_smaps() in my build) and it solves the high
execd load. However, I don`t want to patch the SGE source for every release :)
This whole area is a can of worms. The memory cgroup in recent Linux is
supposed to be the answer, but it has its own problems apart from
finding a clean way to configure all this stuff.
It's worth raising an issue for something like this.
Well, Abaqus seems to perform best if you leave it alone on one or more machines
for itself, so we are just doing that. Therefore we are not that much interested
in accurate memory usage readings in the first place. Licenses are expensive, so
we are trying to optimize throughput in terms of "Abaqus tokens used in least
amount of time".
Regards,
Thomas Mainka
--
Thomas Mainka science+computing ag
System Administration Hagellocher Weg 73
mail: [email protected] 72070 Tuebingen, Germany
tel.: +49 7071 9457 472 www.science-computing.de
--
Vorstandsvorsitzender/Chairman of the board of management:
Gerd-Lothar Leonhart
Vorstand/Board of Management:
Dr. Bernd Finkbeiner, Michael Heinrichs,
Dr. Arno Steitz, Dr. Ingrid Zech
Vorsitzender des Aufsichtsrats/
Chairman of the Supervisory Board:
Philippe Miltin
Sitz/Registered Office: Tuebingen
Registergericht/Registration Court: Stuttgart
Registernummer/Commercial Register No.: HRB 382196
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
