On Wed, Jan 18, 2017 at 06:43:33PM +0000, Wookey wrote: >Package: installation-reports >Severity: grave >Tags: d-i >Justification: renders package unusable > >Dear Maintainer, > >The current installer, with the new 4.9 kernel, is unable to resolve >domains, so is quite seriously broken.
hosts, not domains, but yes... >This was noted during install on an arm64 gigabyte MP30-AR1 >desktop/server, when choose-mirror failed, but it soon became clear >that DNS was not working. > >Testing on an x86 VM with the same daily image (18th Jan 2017) found >the same problem. Going back to the rc1 installer image (4.8 kernel) >it works OK. > >Tests showed that the network came up fine and things are pingable by IP, but >not name: ># ping wookware.org >ping: bad address 'wookware.org' ># ping 93.93.131.118 >PING 93.93.131.118 (93.93.131.118): 56 data bytes >64 bytes from 93.93.131.118: seq=0 ttl=50 time=19.892 ms > >similarly the failing line from the choose-mirror log works if an address is >inserted: >Jan 18 17:04:11 choose-mirror[31201]: DEBUG: command: wget --no-verbose >http://debian-mirror.cambridge.arm.com/debian/dists/stretch/Release -O - | >grep -E '^(Suite|Codename|Architectures):' > >Jan 18 17:04:11 choose-mirror[31201]: WARNING **: mirror does not support the >specified release (stretch) > ># wget --no-verbose >http://debian-mirror.cambridge.arm.com/debian/dists/stretch/Release -O - | >grep -E '^(Suite|Codename|Architectures):' >wget: unable to resolve host address 'debian-mirror.cambridge.arm.com' > ># wget --no-verbose http://10.1.194.51/debian/dists/stretch/Release -O - | gre >p -E '^(Suite|Codename|Architectures):' >Suite: testing >Codename: stretch >Architectures: amd64 arm64 armel armhf i386 mips mips64el mipsel ppc64el s390x >2017-01-18 17:47:12 URL:http://10.1.194.51/debian/dists/stretch/Release >[177979/177979] -> "-" [1] > >resolv.conf is as expected: >search cambridge.arm.com >nameserver 10.1.2.24 >nameserver 10.1.2.23 > >(adding nameserver 8.8.8.8 makes no difference) > >Watching packets go by when doing a VM install it is clear that the >local DNS server returns the correct response, but this is being ignored >or lost by the D-I initrd. For the sake of completeness, I can confirm that the same problem shows up when running on a different network too. I also see this using the oldest amd64 daily I can grab (from 2017-01-17) which has the 4.9 kernel too. I'll see if I can debug this, but I'm not sure where to look straight away. -- Steve McIntyre, Cambridge, UK. st...@einval.com "Managing a volunteer open source project is a lot like herding kittens, except the kittens randomly appear and disappear because they have day jobs." -- Matt Mackall