Date: Sun, 9 Jun 1996 17:53:19 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: taob@io.org (Brian Tao) Cc: freebsd-hackers@freebsd.org Subject: Re: Memory leak or reporting problem in 2.2-960501-SNAP? Message-ID: <199606092253.RAA29425@dyson.iquest.net> In-Reply-To: <Pine.NEB.3.92.960609180042.11543D-100000@zap.io.org> from "Brian Tao" at Jun 9, 96 06:23:22 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > I would expect to see the sum of RSS to exceed actual physical RAM > usage (particularly on our news server, which makes heavy use of > shared memory segments). But to have RSS add up to half the size of > what 'top' says? Something seems broken there. > The RSS refers to the pages mapped into the process, but those pages may or may not be "active." Active has to do with the in-memory priority (sort of). Whether or not a page is mapped has little to do with that. > > The other weirdness I saw this morning was on the abovementioned > news server. I'll just reproduce the output of 'top' and 'ps' here, > since I have *no* idea what was going on here. I restarted innd with > a binary compiled with -g and about half an hour later, the server > swapped itself to death. Note the amount of "active memory" as well > as innd.debug's RSS size. I rebooted the server and now all seems > well, running the same innd.debug binary. > > > # ps aux > USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND > news 23164 6.5 0.2 296 192 ?? D 1:05PM 0:10.49 -zot.io.org HE > root 2 6.0 0.0 0 12 ?? DL Tue12PM 0:13.26 (pagedaemon) > news 22402 4.0 0.2 300 188 ?? D 12:51PM 0:27.33 -zap.io.org HE > root 3 2.3 0.0 0 12 ?? DL Tue12PM 2:31.29 (vmdaemon) > news 23194 2.0 0.0 0 0 ?? Z - 0:00.00 (in.nnrpd) > news 21687 1.7 0.0 24636 12 ?? Ds 12:44PM 3:57.46 (innd.debug) > [...] > <<<<< > EEEK!!! I think that I have a fix for that, that I am willing to commit to -current. 2.2-current is MUCH MUCH better in the VM arena than snap right now. I have been holding off anything but simple bug fixes in current, and I'll commit the proposed (maybe) fix for that tonight (Sun.) It has been very painful, but the VM code is better than it has ever been (except for perhaps a few lurking bugs.) John
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606092253.RAA29425>