On Thu, Jul 07, 2016 at 01:27:03PM +0000, Gianfranco Costamagna wrote: > I still have an yakkety chroot that needs update, and with the following content > > Acquire::http::Pipeline-Depth 0; > #Acquire::http::No-Cache true; > #Acquire::BrokenProxy true; > Debug::Acquire::http "true"; > Debug::pkgAcquire::Worker "true"; > Acquire::Progress::Ignore::ShowErrorText "true"; > > this is the output > I: Copy /etc/apt/apt.conf.d/00aptitude /etc/apt/apt.conf.d/00trustcdrom /etc/apt/apt.conf.d/01-vendor-ubuntu /etc/apt/apt.conf.d/01autoremove /etc/apt/apt.conf.d/01autoremove-kernels /etc/apt/apt.conf.d/10periodic /etc/apt/apt.conf.d/15update-stamp /etc/apt/apt.conf.d/20archive /etc/apt/apt.conf.d/20auto-upgrades /etc/apt/apt.conf.d/20dbus /etc/apt/apt.conf.d/50appstream /etc/apt/apt.conf.d/50unattended-upgrades /etc/apt/apt.conf.d/70debconf /etc/apt/apt.conf.d/99fixbadproxy /etc/apt/apt.conf.d/99synaptic /etc/apt/apt.conf.d/99update-notifier to chroot […] > >That causes apt (or to be specific: apts 'http' method) to display the HTTP > >request it makes as well as the responses it gets while the second shows the > >interprocess communication between apt and its methods (beware, lots of output). > >And third causes apt to tell us the error it ignored, just for completeness. Hmmmm. I see your fixbadproxy config being copied, so I wonder whats up, but its quite obvious that the options didn't take effect at all as, as mentioned (in the quote), the first two are very noisy. > >[aka: No need for wireshark or other "hard" debug stuff just yet, There > >tends to be an option for everything in apt… so much that some are even > >made up and people believe they exist: BrokenProxy from above ;) ] […] > BTW googling for Acquire::BrokenProxy=true shows *lots* of results I know, experts call it "placebo effect". ;) As an APT developer I can certify that this option has no effect because it doesn't exist (anymore). The option used to exist, git history tells me that it was introduced on 1st April 2005 and less than a year later (in Feb 2006, so more than a decade ago!) removed. In case you wonder: signed Release files were a new thing at that point in time and the option forced re-downloading of the Release.gpg file. That was fixed later in a more automatic way & for a few years now apt prefers to use a file called InRelease; only using the pair Release + Release.gpg if it must so even if the code would be still there it wouldn't be reached in your case. What makes this particularily funny is that it is suggested to fix hashsum mismatches which it can't and never could: The error it intended to workaround for the moment was "bad signature on the Release file" (as an old signature in Release.gpg "obviously" doesn't apply to a new Release file). It probably was a bad idea to use such a catchy option name for such a limited purpose code… or perhaps it a very clever april fouls joke – who could tell the difference… Best regards David Kalnischkies
Attachment:
signature.asc
Description: PGP signature