A late reply, but I just did a bunch of Ruby on Rails full stack
profiling with DTrace reading. Here are the references new and old
for anyone who's interested:
<http://blogs.sun.com/bmc/entry/dtrace_on_rails>
Podcast: <http://joyeur.com/2007/05/18/ps-pipe-grep-episode-16-
conference>
<http://blogs.sun.com/bmc/date/20070520>
<http://blogs.sun.com/bmc/entry/dtrace_on_mac_os_x>
<http://dtrace.joyent.com/>
Comment from the JRuby guy: <http://blogs.sun.com/bmc/entry/
dtrace_and_ruby#comment-1159914961000> (who has an interesting recent
interview on theserverside: <http://www.theserverside.com/tt/talks/
videos/CharlesNutterText/interview.tss>)
Quite amazing. I can't wait to try it out.
~Ian
On Mar 5, 2007, at 2:42 PM, Michael Frederick wrote:
I'm in the process of setting up a "real" production environment on
dedicated machines. In the meantime, I'm running everything in
development mode on my desktop. So, that's a local postgresql database
and a WEBrick server launched from radrails.
A major part of my application is migrating data from an old crappy
database to my new spiffy one. I'm using script/runner to do this in a
cronjob. My problem is that in development, it's *really* slow. It
takes about 10 seconds to migrate a single row in the old database to
about 30 rows in the new one. This is all done with simple rails
models, and the vast majority of CPU time is in ruby, not the
database.
Is this performance normal in development mode using WEBrick? Should I
see a huge improvement on the production setup (Apache with fastcgi on
one server, postgres on another)? And what sort of profiling
techniques can I be using to test things, either in development or
production? I'm using the logs, but they're not that useful for the
script/runner.
- Michael
_______________________________________________
Sdruby mailing list
[email protected]
http://lists.sdruby.com/mailman/listinfo/sdruby
_______________________________________________
Sdruby mailing list
[email protected]
http://lists.sdruby.com/mailman/listinfo/sdruby
