[Twisted-Python] Twisted webserver performance issues
leyssw at ihug.co.nz
Sat Nov 7 21:55:49 EST 2009
Thanks for the link info.
I literally just set up Varnish for the first time on my web host, so I
am no pro. However, varnish uses a c like scripting language that can
run regexps on the incoming urls and apply rules to them.
What this means is that we can (as a starting point) choose caching
times for each of those high traffic urls in the varnish script without
touching trac. In the script, we match the front page (with a regexp)
and add a max-cache time to 86400, cache it, etc.
for a great varnish tutorial
By doing that for each of the urls you listed (bar the timeline, which
should be cached for 1 hour or perhaps 30 minutes instead) we will
seriously lighten the load on the database and the webserver, even if
varnish runs on the same machine (as a start).
Every other url (which doesn't have caching set) we could just cache for
30 minutes (perhaps even a day), but only for anon users (no cookies set
>> On 10:41 pm, leyssw at ihug.co.nz wrote:
>>> Isn't the simplest option to place a decent reverse proxy between the
>>> webserver and our clients?
>> Possibly so! I gave this a half-hearted attempted a few weeks ago but
>> quickly gave up. If someone who is familiar with configuring such a
>> proxy would like to help out, perhaps we can get something useful up
> Actually, I just looked at Varnish again and remembered that because of
> the way trac sessions work, it's really hard to figure out what may and
> may not be correctly cached. I don't know how to resolve this without
> serious changes to trac itself.
> Twisted-Python mailing list
> Twisted-Python at twistedmatrix.com
> End of Twisted-Python Digest, Vol 68, Issue 13
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Twisted-Python