GNOME Bugzilla – Bug 777831
nm-online returns success before wlan is connected
Last modified: 2017-02-17 19:44:24 UTC
Created attachment 344418 [details] journalctl -alb with NetworkManager --log-level=TRACE This is related to Debian bug #771441 (https://bugs.debian.org/771441). tftpd is started to early, in it's init.d script it has: # Required-Start: $local_fs $remote_fs $syslog $network and it is really only started once NetworkManager-wait-online.service is complete but still the output of `ip a` is: 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever 2: wlp2s0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default qlen 1000 link/ether 4e:e6:d3:b5:49:da brd ff:ff:ff:ff:ff:ff The problem seems to be that nm-online -s -q --timeout returns before the wlan interface is in connected state.
Just upgraded to network-manager 1.6.0-1, issue still remains
fix for review: https://cgit.freedesktop.org/NetworkManager/NetworkManager/log/?h=th/device-pending-action-bgo777831
I tested the debian packages mbiebl provided: $ dpkg -l network-manager Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend |/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad) ||/ Name Version Architecture Description +++-==============-============-============-================================= ii network-manage 1.6.0-2~test amd64 network management framework (dae I assume that this includes the above patches, but tftpd still fails to start. Best regards Uwe
(In reply to uwe+gnome from comment #3) > I assume that this includes the above patches, but tftpd still fails to > start. I don't know whether the package includes the patches, but if it does and it doesn't fix the issue, a logfile would be helpful.
Version 1.6.0-2~test1 includes the patches from th/device-pending-action-bgo777831 Uwe, can you send a new log file?
Created attachment 344487 [details] journalctl -alb with NetworkManager --log-level=TRACE from th/device-pending-action-bgo777831
branch extended... should fix the issue now. (while at it, I changed several other things, so the resulting branch is rather large)
it's overly 50 patches large, so it's not obvious which specific commits to cherry-pick for nm-1-6 which fixes this bug report. Thomas, could you list the commits I need?
(In reply to Michael Biebl from comment #8) > it's overly 50 patches large, so it's not obvious which specific commits to > cherry-pick for nm-1-6 which fixes this bug report. Thomas, could you list > the commits I need? it's only on-review at this point. I would suggest to wait until the relevant parts are on nm-1-6 branch. Is that ok?
(In reply to Thomas Haller from comment #9) > (In reply to Michael Biebl from comment #8) > > it's overly 50 patches large, so it's not obvious which specific commits to > > cherry-pick for nm-1-6 which fixes this bug report. Thomas, could you list > > the commits I need? > > it's only on-review at this point. I would suggest to wait until the > relevant parts are on nm-1-6 branch. Is that ok? Well, I can certainly wait. I was just trying to prepare a test package for Uwe, so having a minimal patch set would have been nice
Looks generally fine. In general, some commit messages/comments may need improvement: > a7a288b shared: add NM_CACHED_QUARK() macro Missing a comment about what does the macro do. Missing explanation why is it useful in commit message. > a4b4466 core/trivial: use define for GObject property name NM_SUPPLICANT_INTERFACE_SCANNING > 7aef7cf device: add pending-action "recheck-available" > When we have a recheck-available scheduled, we clearly should not yet > declare startup-complete. Not clear to me. Why? > bfb30c7 device: inline device helper structs QueuedState and PingInfo Why?
(In reply to Lubomir Rintel from comment #11) > Looks generally fine. In general, some commit messages/comments may need > improvement: thanks. I improved commit messages of all mentioned.
merged branch to master in 3 parts: 1. https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=61554cae7a48c0b42e28f0893844fa9f17a5d783 2. https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=f3504c9bc214aa32f7a498a1adb48b840be8809a 3. https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=f549307ca5f7a6bf3e925497472616c87c8d4cbe Only part 2 matters for this issue (the other parts are refactoring). Backported part 2 to nm-1-6 as https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=85140df1170b8c0ec3231f59850d5682a99893bf This should fix the issue. Closing.
Created attachment 345481 [details] journalctl -alb with NetworkManager --log-level=TRACE. Debian package for 1.6.0-1 + 256bfe54e7ef..85140df1170b I did something like this to create the package: debcheckout network-manager cd network-manager dquilt push -a dquilt new device-pending-action-bgo777831.patch git diff 256bfe54e7ef 85140df1170b | dquilt fold dquilt refresh dch -l ukl sbuild Then I installed the resulting packages (not all of them, just gir1.2-networkmanager-1.0 libnm-glib4 libnm-util2 libnm0 network-manager)
(In reply to uwe+gnome from comment #14) > Created attachment 345481 [details] > journalctl -alb with NetworkManager --log-level=TRACE. Debian package for > 1.6.0-1 + 256bfe54e7ef..85140df1170b > > I did something like this to create the package: > > debcheckout network-manager > cd network-manager > dquilt push -a > dquilt new device-pending-action-bgo777831.patch > git diff 256bfe54e7ef 85140df1170b | dquilt fold > dquilt refresh > dch -l ukl > sbuild > > Then I installed the resulting packages (not all of them, just > gir1.2-networkmanager-1.0 libnm-glib4 libnm-util2 libnm0 network-manager) A fix for this should be on https://cgit.freedesktop.org/NetworkManager/NetworkManager/log/?h=th/device-wifi-wait-for-scan-bgo770938 for bug 770938. commit: "device/wifi: mark the device to busy while supplicant is in scanning state"
Uwe reported that his problem is still unfixed with 1.6.2, so we should reopen the bug report.
fixed on master: https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=fe6204ffc71f62c1667d0eba56f3a22d6aa9bbbd
nm-1-6: https://cgit.freedesktop.org/NetworkManager/NetworkManager/commit/?id=1dbb6b315072860dfab08766148828d4cd2a44da
I tested Debian-Package 1.6.0-1 + git diff 1.6.0..803e1a04bb0a35b096c1f17bdd0d1abd41f0b863 | filterdiff -p 1 -xcontrib/* and the issue is gone for me. Thanks Uwe