Changes in bug 1394579 are causing regressions in Anaconda which has been assuming one conection per device (the matched ifcfg connection vs additionally generated in-memory connection).
(In reply to Radek Vykydal from comment #0)
> --- Additional comment from Thomas Haller on 2017-04-19 09:33:01 EDT ---
> I see.
> I think we should fix NM so that the heuristic of choosing between a) and b)
> behavior is closer to the previous behavior. Then NM should detect that the
> existing file is a candidate to assume.
we could also set some NM configuration in installer to run in something like the former behavior mode, but perhaps you want to fix it without special casing (ie based just on existence of NM state file?).
*** Bug 1439220 has been marked as a duplicate of this bug. ***
Created attachment 1273895 [details]
logs with new NetworkManager (version 1.8.0-0.4.rc1.el7)
Created attachment 1273896 [details]
logs with new NetworkManager (version 1.8.0-0.2.git20170215.1d40c5f4.el7)
Will this update go to F26 as well? I am not sure if the new build for f26 has it and when it will hit the compose.
In Fedora we have two related bugs (bug 1439388, bug 1439051 - most probably having the same cause). The traceback itself should be fixed by anaconda patch, but we would be probably seeing issues similar to those we've been seeing in RHEL 7 without applying this NM change in Fedora.
(In reply to Radek Vykydal from comment #8)
> Will this update go to F26 as well? I am not sure if the new build for f26
> has it and when it will hit the compose.
Yes. Fedora 26 now has 1.8.0, which has this fix as well. https://bodhi.fedoraproject.org/updates/FEDORA-2017-d285a0c72f
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.