>From: Christian Biere <[EMAIL PROTECTED]>
>Date: Mon, 20 Nov 2006 20:46:52 +0100
>
> > ** ERROR **: mmap(0, 2097152, PROT_READ | PROT_WRITE, 0x22, -1, 0)
>failed:
> > Cannot allocate memory
> > aborting...
>
>We bail out a little sooner than necessary. There might still 16 unused
>MiB.
>This could be fixed by trickling pages down in the page cache.
>
> > This is an obvious error - I ran out of memory. The question is - Why?
>
>The question is whether it used almost all RAM or whether it didn't. You
>could try whether disabling VMM_GREEDY_PAGE_CACHE fixes this for you. There
>could be some memory leak somewhere as well. How many nodes are typically
>connected?
>
It happened again, this time at 1d 10h 30m. Here is some data from my
"cron" watcher:
Wed Nov 22 11:15:01 MST 2006 - TRS:1941, DRS:188194, RSS:127652, %MEM:49.8
Wed Nov 22 11:20:01 MST 2006 - TRS:1941, DRS:178990, RSS:127896, %MEM:49.9
Wed Nov 22 11:25:02 MST 2006 - TRS:1941, DRS:190742, RSS:128292, %MEM:50.1
Wed Nov 22 11:30:01 MST 2006 - TRS:1941, DRS:184910, RSS:127840, %MEM:49.9
Wed Nov 22 11:35:01 MST 2006 - TRS:1941, DRS:178286, RSS:127296, %MEM:49.7
Wed Nov 22 11:40:01 MST 2006 - TRS:1941, DRS:180518, RSS:127172, %MEM:49.6
Wed Nov 22 11:45:01 MST 2006 - TRS:1941, DRS:180226, RSS:127128, %MEM:49.6
Wed Nov 22 11:50:01 MST 2006 - TRS:1941, DRS:178262, RSS:127524, %MEM:49.8
Wed Nov 22 11:55:01 MST 2006 - TRS:1941, DRS:188514, RSS:128144, %MEM:50.0
Wed Nov 22 12:00:01 MST 2006 - TRS:1941, DRS:191266, RSS:128376, %MEM:50.1
I'm not really sure how to interpret this. My understanding of RSS is that
this is the amount of unswappable memory that the process is using, in which
case I had plenty of physical memory to work with (it was only using half of
what was available). The DRS value, however, looks a bit ominous.
I've updated to SVN r12425, and I'm going to recompile with
VMM_GREEDY_PAGE_CACHE disabled to see what the memory signature looks like
without this option. I'll let you know the results in a couple of days.
-------------------------------
Comment: The may be a slow memory leak in GtkG. Here are the memory values
for the entire run, at 1 hour intervals:
Tue Nov 21 02:00:01 MST 2006 - TRS:1941, DRS:47682, RSS:24824, %MEM:9.6
Tue Nov 21 03:00:01 MST 2006 - TRS:1941, DRS:92110, RSS:48336, %MEM:18.8
Tue Nov 21 04:00:01 MST 2006 - TRS:1941, DRS:108102, RSS:65100, %MEM:25.4
Tue Nov 21 05:00:01 MST 2006 - TRS:1941, DRS:112806, RSS:72248, %MEM:28.2
Tue Nov 21 06:00:01 MST 2006 - TRS:1941, DRS:127170, RSS:79000, %MEM:30.8
Tue Nov 21 07:00:01 MST 2006 - TRS:1941, DRS:137042, RSS:85004, %MEM:33.1
Tue Nov 21 08:00:01 MST 2006 - TRS:1941, DRS:142802, RSS:90852, %MEM:35.4
Tue Nov 21 09:00:01 MST 2006 - TRS:1941, DRS:140906, RSS:93500, %MEM:36.5
Tue Nov 21 10:00:01 MST 2006 - TRS:1941, DRS:148582, RSS:96872, %MEM:37.8
Tue Nov 21 11:00:01 MST 2006 - TRS:1941, DRS:154454, RSS:99532, %MEM:38.8
Tue Nov 21 12:00:01 MST 2006 - TRS:1941, DRS:155698, RSS:102300, %MEM:39.9
Tue Nov 21 13:00:01 MST 2006 - TRS:1941, DRS:159418, RSS:101628, %MEM:39.6
Tue Nov 21 14:00:01 MST 2006 - TRS:1941, DRS:156990, RSS:102252, %MEM:39.9
Tue Nov 21 15:00:01 MST 2006 - TRS:1941, DRS:160826, RSS:94608, %MEM:36.9
Tue Nov 21 16:00:01 MST 2006 - TRS:1941, DRS:144138, RSS:94940, %MEM:37.0
Tue Nov 21 17:00:01 MST 2006 - TRS:1941, DRS:147578, RSS:96752, %MEM:37.7
Tue Nov 21 18:00:01 MST 2006 - TRS:1941, DRS:146182, RSS:97380, %MEM:38.0
Tue Nov 21 19:00:01 MST 2006 - TRS:1941, DRS:157062, RSS:99848, %MEM:38.9
Tue Nov 21 20:00:01 MST 2006 - TRS:1941, DRS:149726, RSS:101156, %MEM:39.5
Tue Nov 21 21:00:02 MST 2006 - TRS:1941, DRS:147406, RSS:101400, %MEM:39.6
Tue Nov 21 22:00:01 MST 2006 - TRS:1941, DRS:146794, RSS:98876, %MEM:38.6
Tue Nov 21 23:00:01 MST 2006 - TRS:1941, DRS:150626, RSS:102312, %MEM:39.9
Wed Nov 22 00:00:01 MST 2006 - TRS:1941, DRS:143902, RSS:100160, %MEM:39.1
Wed Nov 22 01:00:01 MST 2006 - TRS:1941, DRS:151842, RSS:98912, %MEM:38.6
Wed Nov 22 02:00:01 MST 2006 - TRS:1941, DRS:164334, RSS:106932, %MEM:41.7
Wed Nov 22 03:00:01 MST 2006 - TRS:1941, DRS:164662, RSS:107996, %MEM:42.1
Wed Nov 22 04:00:01 MST 2006 - TRS:1941, DRS:160494, RSS:110424, %MEM:43.1
Wed Nov 22 05:00:01 MST 2006 - TRS:1941, DRS:169722, RSS:112608, %MEM:43.9
Wed Nov 22 06:00:01 MST 2006 - TRS:1941, DRS:176890, RSS:115688, %MEM:45.1
Wed Nov 22 07:00:02 MST 2006 - TRS:1941, DRS:178614, RSS:117896, %MEM:46.0
Wed Nov 22 08:00:01 MST 2006 - TRS:1941, DRS:178574, RSS:117800, %MEM:46.0
Wed Nov 22 09:00:01 MST 2006 - TRS:1941, DRS:172062, RSS:119420, %MEM:46.6
Wed Nov 22 10:00:01 MST 2006 - TRS:1941, DRS:180086, RSS:127224, %MEM:49.6
Wed Nov 22 11:00:01 MST 2006 - TRS:1941, DRS:184830, RSS:128188, %MEM:50.0
Wed Nov 22 12:00:01 MST 2006 - TRS:1941, DRS:191266, RSS:128376, %MEM:50.1
I would have expected the memory usage to stabilize after 12 hours or so as
an ultra, but it never does. Though the graph is far from smooth, the trend
is generally upwards for almost the entire run. This worries me...
If anyone is interested, I've retained the data (at 5 minute intervals). I
didn't want to post that much to the list.
Lloyd Bryant
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Gtk-gnutella-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel
