Server now running lighttpd

[SpankMe]

In an effort to delay having to upgrade the server I have replaced Apache with lighttpd (http://www.lighttpd.net/). This web server software uses less resources and is supposed to give better response times than Apache. lighttpd powers web sites like YouTube & wikipedia.

As usual if you encounter any problems, please let me know here.

[limbimtimwim]

Have you tried a php compiler thing like turck mmcache? Pretty amazing speed ups of php pages. I've always been scared it would break things, but I've never seen it break anything.

[SpankMe]

I've been using APC (http://au3.php.net/apc) for awhile.

[Kendog]

KB has been running slowly, seems to be off and on. But when it slows down it is very noticeable.

[nadroj]

I had trouble logging in since about then. It accepts my name & password + welcomes me but returns to the login screen. Have I been naughty? Not on here. I fluked acces this morning via register page.
Cheers

[Flyingpony]

Was checking my emailed quote notifications late last night without being logged into KB.
Could read most of them but some threads asked me to log in :scratch:
Namely the "Haldanes's Motorcycles is moving" one.
Don't know if there means there's a problem or not, just thought I'd share it.

[nadroj]

I had trouble logging in since about then. It accepts my name & password + welcomes me but returns to the login screen. Have I been naughty? Not on here. I fluked acces this morning via register page.
Cheers

Ran ad-aware & now OK

[limbimtimwim]

I've been using APC (http://au3.php.net/apc) for awhile.Nice!

I see you have gone back to Apache 2.0.x

Are you sure the webserver is the bottleneck? It is not the DB? Spawning new httpds is often the most costly operation. To slow down httpd creation see if check that MaxRequestsPerChild is high enough? MinSpareServers is a reasonable number?

Perhaps use the worker mpm instead of the prefork mpm. I've never used worker, so good luck with that if you do :)

Perhaps put the CSS in a seperate regular file browsers can cache? I don't know anything about web design, so that might be a no-no.

Just putting the low hanging fruit out there..

[Fatjim]

Mate, you haven't got any low hanging fruit!

It's obviously the thrombibulator is turned up to high.

[RantyDave]

Perhaps put the CSS in a seperate regular file browsers can cache?
I don't think that's an option because it would mean hacking around with vBulletin - which is far too much like hard work.

I'm just trying to find the thing that looks at disk performance ... the server may be short on RAM and hence just stuffing another stick in would solve the problem for a while. Also, PostgreSQL databases need vacuuming from time to time - maybe it needs a bit of a clean up?

Ah. "iostat -d" part of the sysstat package (on debian).

Dave

[Drum]

No real problems, but definitely seems to be running slower.

[Korea]

I don't think that's an option because it would mean hacking around with vBulletin - which is far too much like hard work.

I'm just trying to find the thing that looks at disk performance ... the server may be short on RAM and hence just stuffing another stick in would solve the problem for a while. Also, PostgreSQL databases need vacuuming from time to time - maybe it needs a bit of a clean up?

Ah. "iostat -d" part of the sysstat package (on debian).

Dave

Way too clever for me, kids.
Um, yeah... really enjoying this KB thing. Cheers. :rockon:

[SpankMe]

Some software I use stoppped working under lighttpd so it's back to apache until I can work out the bugs on my dev box.

[limbimtimwim]

Also, PostgreSQL databases need vacuuming from time to time - maybe it needs a bit of a clean up?
Ah. "iostat -d" part of the sysstat package (on debian)vB uses pgsql?

pgsql vacuums only are required if you are deleting rows. You need to actually 'free' the row after it has been deleted. Don't think KB does much row deletion.