[dokuwiki] Re: Migration & Plugin for MoinMoin

  • From: Yann <yann.hamon@xxxxxxxxx>
  • To: dokuwiki@xxxxxxxxxxxxx
  • Date: Wed, 24 May 2006 15:18:35 +0200

What performance problems are you having?

I'm having heavy load due to what I think is a bug in the changes.logparsing. Once or twice a week, the load inscreases up to 20 (see http://isfates.mandragor.org/toto/dokuwikibug.png ) ... emptying the changes.log and killing php4 processes solves the problem. See the thread "Performances issues" from march 12. I proposed a patch, which is not optimal; and I don't have the time to code the whole changes.log parsing again.

- can't handle the number of visitors?
- can't handle the volume of page edits?


I think that's the problem.


- can't handle the size/complexity of your pages?
- do your pages have lots of code blocks with language highlighting (by
GeSHi)?


No.


- can't handle the number of RSS feed requests?


There is a huge performance problem in an utf8 function in the RSS code; see the topic "performance issues", Mario Emmanlauer, march 25.


- something else?

How many pages does your wiki have?


yann@mawu:/srv/www/fr/doc.ubuntu-fr.org/htdocs/data/pages$ find -iname
"*.txt" | wc -l
1114

What is there average size?


Well, "normal"? There are not really long pages afaik.


How often do they get edited?


Let's say there are 50 to 100 page edits per day..


Do your page make heavy use of images?

Not really.

How many visitors, page visits do you get?


http://master.ubuntu-eu.org/weblogs/doc.ubuntu-fr.org/usage_200605.html


Do many people link to the wiki's RSS feed - did you upgrade to the
recent patches which made improvements to the RSS feed?


The RSS feed has been chmoded 000.


What OS/Webserver do you run?


Ubuntu server, on a 2*1GHZ. pentium, 1GB RAM.

Do you have page compression (DEFLATE on apache) turned on?


We don't.


So, as I said, the main problem comes of the parsing of the changelogs. The way it is actually done _is_ really, really bad, and just is not possible anymore with a really heavy changes.log. Take the example of a changes.logwhich is 5mbyte big. Take the example of a page that has been edited 150 times. The code "latest edits" will do 150 greps on that 5mbyte file, to get the list of the modifications. Now just imagine, someone spidering the website; until now we have had approximatively 24000 page édits; that means, if he downloads all the "last edits" pages, so for 1000 pages, it'll be more than 24000 grep on a 5Mbyte file. Im not quite surprised the load becomes so high...

Then there has been the RSS issues, and the antispam feature which is quite
heavy too.

There's still the fact that I don't understand why I need to empty the
changes.log and to kill the php4 processes to get the load back to normal...
And that I don't understand why the load becomes so high all of a sudden.
But I'm tired of having to empty that changes.log file every 3-7 days.

Cheers,
Yann

Other related posts: