Date: Thu, 20 Nov 1997 21:41:10 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: mrcpu@cdsnet.net (Jaye Mathisen) Cc: toor@dyson.iquest.net, hackers@freebsd.org Subject: Re: Nope, didn't fix it.Re: Serious performance issue with 2.2.5-RELEASE Message-ID: <199711210241.VAA00867@dyson.iquest.net> In-Reply-To: <Pine.NEB.3.95.971120170004.23652a-100000@mail.cdsnet.net> from Jaye Mathisen at "Nov 20, 97 05:05:49 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Jaye Mathisen said: > > OK, I handpatched the source, built a new kernel, and fired it up again. > > If anything, it's worse than before, hanging up within seconds of > rebooting. > Hmmmm... The original modification of the code was a mistake. getnewbuf is supposed to return if it blocks. > > I'm dropping back to the previous version for now, but this pretty well > kills 2.2.5 for me for usability. > Ouch! At least you have a plan B. > > I notice now that a lot of processes seem to be in vmwait, when they > didn't used to show it. > > Heck, I just had a la spike of 480. Machine kept running, but it was > unable to serve files for 2-3 minutes. > Whoa!!! Bring in the vm_page fix from RELENG_2_2 !!! It could be another manifestation of the page queue management bug!!! John
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199711210241.VAA00867>