On Mon, 2006-07-17 at 00:15 +0200, Henrik Nordstrom wrote: > sön 2006-07-16 klockan 21:49 +0800 skrev Adrian Chadd: > > > I've written some code to push logfile writing into an external process, > > freeing up the main squid process from the potentially blocking > > stdio writes. > > Good. > > > This code isn't ready to be merged into squid-2.6. Its meant more as a > > prototype. I haven't even reviewed it for 'correctness' besides running > > it under a reasonable load (200req/sec) for half a day. > > Some comments: > > Instead of the linked list of separately maintained buffers, have you > considered using MemBuf:s? > > > * rework the socket IO scheduling code to try and hold onto a buffer until > > its full or 1 second has passed. It might not really matter but it'll be > > interesting to profile it at very high (>1500/sec) request rates. > > Not sure I agree with this. It should even out anyway under load, and > under low load it's just annoying to have logs delayed. > > > * figure out what to do about rotate (as it stands, this code only re-opens > > the logfiles and therefore pushes logfile maintainence into an external > > script.) > > You flush any pending buffers first, right?
I did a patch for squid 2.x/early 3.0 that did something similar - and it suffered from dropped log contents during high load... which is something I never got *satisfactorily* addressed - its the key thing I think needs considering. Rob -- GPG key available at: <http://www.robertcollins.net/keys.txt>.
signature.asc
Description: This is a digitally signed message part
