Sometimes you should never jump to conclusions before looking at the evidence. I requested a list of scripts which were causing problems, and this was the result:
| process | count | CPU secs | CPU minutes |
|---|---|---|---|
| php | 14871 | 5656.85s | 94.28 CPU minutes |
| mt-postacomment | 39 | 27.70s | 0.46 CPU minutes |
| awstats.pl | 8 | 11.20s | 0.19 CPU minutes |
| imapd | 55 | 8.81s | 0.15 CPU minutes |
| mt-feed.cgi | 8 | 7.00s | 0.12 CPU minutes |
So, as you can see, it’s not Movable Type’s comments after all – in 1 day, it was using a mere 30 seconds of CPU time, compared with an hour and a half for PHP. Now ‘php’ includes all PHP scripts on the site, which does include parts of Movable Type – namely the dynamically generated pages and the search results. But there are other scripts on here – my own tag search script and some bot-banning scripts. I have disabled these other scripts, and reduced the number of pages being served by dynamic publishing, to see if this makes any difference. Hopefully I will find out soon.
So this means I won’t be switching to WordPress right away. I will keep it on here for now, in case I want to take a look at it, but for now I’ll stick with Movable Type. I will re-enable comments, and I have brought over the entries and discussions from the temporary WordPress blog so that they’re here to view.
April 5, 2008 at 20:41
As with trying to find slowdowns in my programs, nothing beats cold hard data before attacking the real problem rather than the “obvious” one…
Regards,
Rob…
June 12, 2008 at 15:16
No, it’s ok. I’ve found that MT is pretty much a steaming pile. WP can be considered a bit clunky as well, but not nearly that pita that MT is.