On Fri, 2007-11-30 at 10:59 -0800, Ross Boylan wrote: > On Fri, 2007-11-30 at 11:44 +0000, Mark Hindley wrote: > > All very interesting. It really doesn't look like apt-cacher is the > > problem here. Certainly the patch I posted should make apt-cacher play > > more nicely with upstream caches. But it can't force them to do the > > right thing. > I've had no problems since applying the patch, and I think usually I > would have. But I'd like to try perhaps another week to be sure. > > > > Are you happy this is not an issue I need to address? I have been > > holding off uploading 1.6.0 to try to get this fixed. > It looks as if some of the upstream servers are misbehaving. However, > since I've had the problem with several of them, it would be good if > apt-cacher could cope with the misbehavior. I also wonder if this is > exposing some misbehavior in the Debian archive (server) software. > > I think I'll write the admin of the the server with the recent problem. > They might have more insight. Or them might ignore the message! > > Ross I just got a GPG verify error against security.debian.org, testing/updates. I tried twice and failed; the 3rd time succeeded.
On the failure, the logs show HTTP giving timestamps of 08 Dec 16:51 GMT Release 08 Dec 17:26 GMT Release.gpg. The fetches occurred at 17:31 and 17:42 GMT. I used lftp to examine the site, and saw that both files had timestamps of 17:43. (So the Release key was regenerated after 17 minutes?!) The next update succeeded, and the apt-cacher logs show HTTP giving 17:43 as the timestamp for both files. So this seems to have been caused by catching the archive in mid-update. Yesterday I got a GPG error on a machine use apt without apt-cacher; I'm not sure if that was against security or the regular mirror. At any rate, apt-cacher seems to be producing these errors no more frequently than other methods. Ross -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
