importd servers don't deal with invalid certificates gracefully

[Tom Haddon]

The following output is seen on https://code.edge.launchpad.net/~vcs-imports/rdfbean/trunk:

- Hostname: source.mysema.com
 - Valid: from Mon, 18 Aug 2008 10:49:13 GMT until Thu, 16 Aug 2018 10:49:13 GMT
 - Issuer: IT, Mysema Ltd, Helsinki, Helsinki, FI
 - Fingerprint: a1:1f:62:3c:39:2a:53:34:79:3d:45:7c:87:e2:3b:b7:af:73:e1:86
(R)eject, accept (t)emporarily or accept (p)ermanently?

This means it's going to hang waiting for a response. The issue in question is a self-signed certificate.

[Aaron Bentley (abentley)]

It appears that the SSL certificate has been added manually to russkaya and neumayer, but somehow galapagos was missed. Hopefully, we can improve this situation now that we're using bzr-svn.

[Tom Haddon (mthaddon)]

Two more - https://code.edge.launchpad.net/~vcs-imports/mpd/trunk and https://code.edge.launchpad.net/~vcs-imports/iptables/main

[Tom Haddon (mthaddon)]

Because galapagos was re-installed recently, all of the "remember this cert as being okay" answers would have been forgotten. Clearly, that isn't a good way of managing that.

[Michael Hudson-Doyle (mwhudson)]

Aaron and I figured out how to make this work gracefully with bzr-svn and bzr-hg, so at least we can solve the problems for new imports...

[Michael Hudson-Doyle (mwhudson)]

The fix is churning it's way through EC2 now, and has been tested to work on staging already.