[CentOS] yum vs up2date

Les Mikesell lesmikesell at gmail.com
Thu Sep 7 23:09:28 UTC 2006

On Thu, 2006-09-07 at 22:52 +0100, Lance Davis wrote:

> > That would also be fine if it always gave the same list in the same
> > order when asked through the same proxy -
> It will until it next checks the mirrors - which happen approx hourly - 
> and then the mirrors within a cc are used randomly so as not to overload 
> any particular mirror with al the traffic.
> > if the client walks the list
> > in order.
> It does - unless you use fastestmirror plugin
> > Or if the URL to get this list was always the same and
> > it is marked cachable for some time.
> It is always the same - mabe it needs marking as cachable ??

Hmmm, I think the default squid config excludes caching urls with
? argument lists - which usually makes sense.  Could you, instead
of randomizing the list at time intervals, construct 10 versions
in differently sorted orders every time with a technique that
will make each always the same unless the valid servers change,
then return one of these versions based on a hash of the requesting
IP address to make the distribution random but always give the
same file version to the same requester?  Then that file should
always have the same first choice unless your probe took it out
of the list.  

> > You don't have to control a machine to give out it's IP in a DNS
> > response. You only have to arrange for it accept the name as a vhost
> > and map the document root to the top of your mirror tree. That still
> > may be a lot to ask but it is a very different question.
> It is more than most mirrors will do - especially when they have varying 
> format urls

Agreed - I suspect you could find volunteers that would do it if you
asked for a 'virtual' server instead of hardware, but intelligent
mirror list management is probably better all around.

> > The nature of your product is such that the odds are good that vast
> > numbers of those boxes are located in some small number of places that
> > would only pull one copy of an update if you didn't go out of your way
> > to force each machine to get its own.
> But at least we get to see how many machines are out there :)

If you don't make the mirrorlist cachable you'll still get a good
count of requests although you probably can't distinguish machines
behind proxies. Making it (usually) return the same first choice
to the same proxy requester will fix the rest.

  Les Mikesell
   lesmikesell at gmail.com

More information about the CentOS mailing list