Bug#442189: race condition
As Eugugene stated, it's because of a race condition.
Another related bug, #439031, indicated that multiple gpgv threads run on
the same Releases files (once for a deb, again for deb-src line in
sources.list) causes a similar stall.
I've made a trivial-style repo from dpkg-scanpages (results piped though
plain gzip) and set a chrooted etch to download directly from it. It
stalled on apt-get update. So that ruled out the http-proxy and
mini-dinstall:python:BZ2 issues I was suspecting.
What _did_ help was only using a deb-src line against that repo. And I bet
that my prior report of success with my new reprepro repositiory was
beacuse reprepro uses a non-trivial layout (multiple Packages files).
I guess chroots just trigger the race easily for some reason, why else the
lack of reportage/fix-urgency?
joey
Reply to: