On Thu, Dec 07, 2006 at 09:31:36AM +0000, Justin Mason wrote: > > and got freshclam to run as a daemon - so it > > could randomly sleep between lookups - and thus spread the load. > > I can think of a useful modification -- change sa-update so that, if it's > run non-interactively, it sleeps for a random amount of 0-600 seconds. > That would reduce the hit.
I'm not sure how this would help exactly. If people want to check for updates
once a minute, and sa-update sleeps randomly for up to 10m, this just means
their system will have (potentially) 10 sa-updates running at the same time.
Through the magic of randomness, all 10 of those could end up making requests
at the same time (or at least within the same minute), and that's a lot less
spread out than once a minute.
> However note that we also support any number of mirror servers, too.
> given that, I think it's doubtful we're going to run into this
> problem...
Yeah, the design, I think, is pretty scalable.
--
Randomly Selected Tagline:
"Don't ever make trouble here, I beat you up each time."
- From Rumble in the Bronx
pgpVllgCOSCT6.pgp
Description: PGP signature
