Description of problem: Network was taking about 1 min 30 secs to come up. 45 secs each loopback and eth0 Version-Release number of selected component (if applicable): systemd-16-1.fc15.x86_64 BUT, probably an earlier version. How reproducible: every time Steps to Reproduce: 1. boot system, get a cup of coffee while network comes up 2. 3. Actual results: working system, long boot time Expected results: no delay in starting network Additional info: found both network.service and NetworkManager.service enabled in systemd. Disabling one or the other results in normal boot with network coming up without delay. I don't know at what point both became enabled. systemd was installed at some point last year when it became the default for rawhide, but the network activation delay did not occur until some months later.
I suspect this is the same as https://bugzilla.redhat.com/show_bug.cgi?id=659647 . -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
are you sure it works normally if you enable network but not NetworkManager? -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
I can reproduce this with these steps: run system with configured network service and without installed NetworkManager - e.g. minimal installation # yum install NetworkManager # systemctl | grep -i network network.service loaded active exited LSB: Bring up/down network network.target loaded active active Network # service network restart Restarting network (via systemctl): [ OK ] and you can see in logs: [ 284.080826] systemd[1]: Trying to enqueue job NetworkManager.service/start/replace [ 284.082789] systemd[1]: Installed new job NetworkManager.service/start as 184 [ 284.084302] systemd[1]: Installed new job dbus.target/start as 185 [ 284.085702] systemd[1]: Enqueued job NetworkManager.service/start as 184 [ 284.087102] systemd[1]: dbus.target changed dead -> active [ 284.088131] systemd[1]: Job dbus.target/start finished, success=yes [ 284.089445] systemd[1]: About to execute: /usr/sbin/NetworkManager --no-daemon [ 284.096874] systemd[1]: Forked /usr/sbin/NetworkManager as 1113 [ 284.106669] systemd[1]: NetworkManager.service changed dead -> start # systemctl | grep -i network network.service loaded active exited LSB: Bring up/down network NetworkManager.service loaded active running Network Manager network.target loaded active active Network
(In reply to comment #2) > are you sure it works normally if you enable network but not NetworkManager? > -- > Fedora Bugzappers volunteer triage team > https://fedoraproject.org/wiki/BugZappers Yes. Very sure. To resolve initially I did the following (approx. since my memory is faulty): 1. system-config-network to turn off NetworkManager managing the connection 2. systemctl --disable NetworkManager.service 3. systemctl --enable network.service (did not have to configure, since it was already configured for eth0) 4. reboot and all was well. Based on msgs on the test list I later disabled network.service, enabled NetworkManager.service and configured the interface to let NetworkManager control the interface, rebooted and all was well.
According to Dan the network script and NM should actually be running at the same time. NM checks whether an interface is configured via the old network script before controlling it itself.
On current F15 NM and the network script should work fine together. Closing. If not, feel free to reopen.
*** Bug 659647 has been marked as a duplicate of this bug. ***