On Wed, 2 Nov 2005, Peter Cordes wrote: > On Mon, Oct 31, 2005 at 12:00:10PM +0100, Dag Wieers wrote: > > On Sat, 29 Oct 2005, Peter Cordes wrote: > > > > > Package: dstat > > > Version: 0.6.1-1 > > > Severity: normal > > > > > > When I leave dstat running it will be using 350MB of memory after a > > > day or so. It will keep using more until the kernel's out-of-memory > > > killer gets it. I usually start dstat with a command line like > > > dstat -D hda,sda,hdc,sdb 20, and I run it with TERM=screen (inside a > > > screen session, of course). > > > > > > I have python-psyco installed, in case that matters. I _think_ I've > > > seen this bug on systems where I didn't have python-psyco installed, > > > though. > > > > You are the second person to report it. Can you test if it only happens > > with the 'Disk' stats ? > > > > dstat -d -D hda,sda,hdc,sdb 20 > > > > I haven't looked at it myself though. Need to look into how to print all > > the data-structures/sizes to see what exactly is increasing. > > > > Thanks for reporting ! > > > I started three copies of dstat running in three screen windows on Oct > 31st. (So they've been running for 1 day and 19 hours.) > $ ps -C python -O etime,pmem,rss,size,vsz -w > PID ELAPSED %MEM RSS SZ VSZ S TTY TIME COMMAND > 8345 1-19:54:02 20.9 326184 398916 401832 S pts/4 00:02:37 python > /usr/bin/dstat -d -D hda,sda,hdc,sdb 20 > 10307 1-18:53:29 20.4 318172 391028 393944 S pts/6 00:02:34 python > /usr/bin/dstat -cg 20 > 10378 1-18:50:03 20.4 317660 390484 393400 S pts/5 00:01:39 python > /usr/bin/dstat -ny 20 > > I have another long-running python process (bittornado) which hasn't > ballooned up, and dstat didn't used to do this.
dstat doesn't do this on my system though. And I'm suspecting either the python release or something specific to your system that influences a decision. (kernel, terminal settings, ...) Can you send me the output of dstat --version and send me the output ? That way I can at least verify that information with the other person. Kind regards, -- dag wieers, [EMAIL PROTECTED], http://dag.wieers.com/ -- [all I want is a warm bed and a kind word and unlimited power] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
