27.09.2015 10:29, Felix Miata пишет:
Andrei Borzenkov composed on 2015-09-27 09:40 (UTC+0300):
Felix Miata composed:
once 210 is installed, the reboot command, if noauto NFS(4) fstab entries have been manually started, results in intolerable delay between reboot command start and restart success.
Your systemd configuration dump shows two issues
a) network-online.target is not started. The end effect is lack of serialization between NFS unmount and stopping of interfaces on shutdown. When NFS mount is started automatically from fstab, network-online.target is pulled in, but not when NFS is mounted manually. I believe there were some patches related to this but I cannot find anything looking now. The simplest fix is to manually enable network-online.target:
ln -s /usr/lib/systemd/system/network-online.target /etc/systemd/system/default.target.wants.
Booted host hs80e with systemd-210 already installed to try this. Ran that command only, then rebooted, then mounted nfs mounts, then tried reboot command. It still takes more than 6 minutes to reboot.
This problem is more or less openSUSE 13.1 specific, because in subsequent releases (and other distros I am aware of) interfaces are started before network.target which is normally always active. May be I had to use network.target instead too in my patch.
b) network@.service lacks dependency on network-online.target. This should have been fixed in bnc#857031. Did you miss some updates?
If it only happened on one installation I might imagine that possible, but this is a problem on all the at least half dozen installations I've installed systemd-210 on as followup to https://bugzilla.opensuse.org/show_bug.cgi?id=924488 and https://bugzilla.opensuse.org/show_bug.cgi?id=832220 . All such installations have had zypper up run on them many times since https://bugzilla.opensuse.org/show_bug.cgi?id=857031 was marked fixed 19 months ago.
Could you paste /usr/lib/systemd/system/network@.service?
I guess for the time being until a fix is found for this I can use .bashrc's Reboot[1] instead of system's reboot on the test systems that use the alias to execute manual nfs mounts, if I can remember needing it.
Doesn't http://download.opensuse.org/repositories/home:/tsaupe:/branches:/openSUSE:/... include a good enough fix to include in updates?
The problem is not in systemd, but in services used to configure interfaces which are from openSUSE. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org