[Yum-devel] speeding up yum was Re: [Bug 207] yumError.py patch submissio

Michael Stenner mstenner at linux.duke.edu
Thu Feb 26 16:55:18 UTC 2004


On Thu, Feb 26, 2004 at 10:48:46AM -0600, David Farning wrote:
> On Thu, 2004-02-26 at 08:53 -0700, Michael Stenner wrote: 
> > The threading (or forking) idea is coming up a lot lately, so I'm
> > beginning to consider it more seriously.  However, I don't understand
> > how threading would help in this case.  It's a single file.  Can you
> > explain what you mean a bit more?
> 
> The situations where threading is useful is where there are several
> files to files to pull at the same time.  ie when you are pulling
> headers from several different repos.

Sure.  I'll buy that.  You can consider that addressed in the other
email I just sent, too :)

> In this case it looks like a good bit of time is wasted establishing the
> connection with the livna servers.  These tests are being run in c using
> libcurl but the results should be comparable to those in pthon;)

libcurl is and will be faster than URLGrabber, but the relative
improvement should be similar.

> hmm... Forgot about failover.  To keep the interface as clean as
> possible, I assume that all the needed files are known ahead of time and
> passed as a list.  Dealing with failover makes things a good bit more
> ugly. 

Not sure I follow you here.  It wouldn't be too bad (technically) to
do each repo in a single thread.  They would be largely independant of
each other.  Failover would occur in each thread, but there'd be no
need to cross threads.

					-Michael
-- 
  Michael D. Stenner                            mstenner at ece.arizona.edu
  ECE Department, the University of Arizona                 520-626-1619
  1230 E. Speedway Blvd., Tucson, AZ 85721-0104                 ECE 524G



More information about the Yum-devel mailing list