[Users] host deploy and after reboot not responsive

Tue Jan 22 23:50:59 UTC 2013

On Tue, Jan 22, 2013 at 9:02 PM, Dan Kenigsberg wrote:
> That's great, as it reduces the question to: why doesn't ntpd start on
> your machine after boot. Do you have any guess? and ntpd logs to share?
> Any peculiar ntp.conf setting?

Hum... I think in Fedora 18 itself and/or due to oVirt setup there is
some conflict regarding network synchronization services....

In fact together with standard ntp there is chrony that I didn't know
until today...

My suspect is this one:

- At install time Fedora 18 by default installs now chrony and doesn't
install instead ntpd

In fact my anaconda log in /root
[root at f18ovn03 ~]# ll /root/anaconda-tb-38wvGO
-rw-r--r--. 1 root root 1493714 Dec 18 14:27 /root/anaconda-tb-38wvGO
contains:

14:19:00,759 DEBUG packaging: select package chrony

and no entry for ntp

At this moment chronyd was still enabled as per default at install time:
[root at f18ovn03 ~]# systemctl status chronyd.service
chronyd.service - NTP client/server
 Loaded: loaded (/usr/lib/systemd/system/chronyd.service; enabled)
 Active: inactive (dead) since Tue, 2013-01-22 14:15:42 CET; 9h ago
Main PID: 1348 (code=exited, status=0/SUCCESS)
 CGroup: name=systemd:/system/chronyd.service

Jan 22 14:15:40 f18ovn03.ceda.polimi.it chronyd[1348]: chronyd version
1.27-git1ca844 starting
Jan 22 14:15:40 f18ovn03.ceda.polimi.it chronyd[1348]: Linux kernel
major=3 minor=6 patch=11
Jan 22 14:15:40 f18ovn03.ceda.polimi.it chronyd[1348]: hz=100
shift_hz=7 freq_scale=1.00000000 nominal_tick=10000...ll=2
Jan 22 14:15:41 f18ovn03.ceda.polimi.it systemd[1]: Started NTP client/server.
Jan 22 14:15:42 f18ovn03.ceda.polimi.it systemd[1]: Stopping NTP
client/server...
Jan 22 14:15:42 f18ovn03.ceda.polimi.it systemd[1]: Stopped NTP client/server.

- When I ran install of ovirt-engine it pulled in ntp as one of its
dependencies, enabling it
In yum.log in fact I find:

Jan 15 06:05:53 Installed: libvirt-0.10.2.2-3.fc18.x86_64
Jan 15 06:05:53 Installed: mom-0.3.0-1.fc18.noarch
Jan 15 06:05:54 Installed: fence-agents-3.1.10-1.fc18.x86_64
Jan 15 06:05:54 Installed: usbredir-0.6-1.fc18.x86_64
Jan 15 06:05:54 Installed: ipxe-roms-qemu-20120328-2.gitaac9718.fc18.noarch
Jan 15 06:05:55 Installed: 2:qemu-system-x86-1.2.2-1.fc18.x86_64
Jan 15 06:05:55 Installed: 2:qemu-kvm-1.2.2-1.fc18.x86_64
Jan 15 06:05:56 Installed: autogen-libopts-5.12-2.fc17.x86_64
---> Jan 15 06:05:56 Installed: ntp-4.2.6p5-8.fc18.x86_64  <------
Jan 15 06:06:02 Installed: vdsm-4.10.3-0.78.gitb005b54.fc18.x86_64
Jan 15 06:06:04 Installed: tuned-2.1.2-1.fc18.noarch
Jan 15 06:06:04 Installed: vdsm-cli-4.10.3-0.78.gitb005b54.fc18.noarch
Jan 15 06:06:04 Installed: 2:qemu-kvm-tools-1.2.2-1.fc18.x86_64

Probably chronyd runs before ntpd and creates conflict.....

So I have tested disabling chrony
# systemctl disable chronyd

and re-enabling ntp as a dependency for vdsmd to see if something changes:
Changed /usr/lib/systemd/system/vdsmd.service matching:
Requires=multipathd.service libvirtd.service ntpd.service

Bingo!

After reboot
[root at f18ovn03 ~]# systemctl status ntpd.service
ntpd.service - Network Time Service
 Loaded: loaded (/usr/lib/systemd/system/ntpd.service; enabled)
 Active: active (running) since Wed, 2013-01-23 00:31:00 CET; 1min 47s ago
Main PID: 1424 (ntpd)
 CGroup: name=systemd:/system/ntpd.service
 └ 1424 /usr/sbin/ntpd -u ntp:ntp -g

Jan 23 00:31:07 f18ovn03 ntpd[1424]: Listen normally on 8 em3
fe80::21c:c4ff:feab:3add UDP 123
Jan 23 00:31:07 f18ovn03. ntpd[1424]: Listen normally on 9 em2
fe80::21e:bff:fe21:b8c6 UDP 123
Jan 23 00:31:07 f18ovn03 ntpd[1424]: Listen normally on 10 em1
fe80::21e:bff:fe21:b8c4 UDP 123
Jan 23 00:31:07 f18ovn03 ntpd[1424]: peers refreshed
...

[root at f18ovn03 ~]# systemctl status vdsmd.service
vdsmd.service - Virtual Desktop Server Manager
 Loaded: loaded (/usr/lib/systemd/system/vdsmd.service; enabled)
 Active: active (running) since Wed, 2013-01-23 00:31:10 CET; 2min 25s ago
Process: 1434 ExecStart=/lib/systemd/systemd-vdsmd start (code=exited,
status=0/SUCCESS)
Main PID: 2778 (respawn)
 CGroup: name=systemd:/system/vdsmd.service
 ├ 2778 /bin/bash -e /usr/share/vdsm/respawn --minlifetime 10 --daemon
--masterpid /var/run/vdsm/r...
 ├ 2781 /usr/bin/python /usr/share/vdsm/vdsm
 ├ 2801 /usr/bin/sudo -n /usr/bin/python
/usr/share/vdsm/supervdsmServer.py 95a1f3e1-8c0d-4efe-a65...
 ├ 2802 /usr/bin/python /usr/share/vdsm/supervdsmServer.py
95a1f3e1-8c0d-4efe-a65a-c5af1c7f4d61 27...
 ├ 3035 rpc.statd --no-notify
 ├ 3041 /usr/bin/python /usr/share/vdsm/storage/remoteFileHandler.pyc 31 30
 ├ 3214 /usr/bin/python /usr/share/vdsm/storage/remoteFileHandler.pyc 44 43
 ├ 3216 /usr/bin/python /usr/share/vdsm/storage/remoteFileHandler.pyc 49 46
 └ 3229 /usr/bin/python /usr/share/vdsm/storage/remoteFileHandler.pyc 37 34

Jan 23 00:31:38 f18ovn03 rpc.statd[3035]: Version 1.2.7 starting
Jan 23 00:31:38 f18ovn03 rpc.statd[3035]: Flags: TI-RPC
...

I also modified ntp.conf commenting out
#server 0.fedora.pool.ntp.org iburst
#server 1.fedora.pool.ntp.org iburst
#server 2.fedora.pool.ntp.org iburst
#server 3.fedora.pool.ntp.org iburst

and putting a "server xxx.yyy.www.zzz" line to an internal ntp server
but I don't think this was influencing my problems
To be sure anyway I reverted to old ntp.conf config with
server 0.fedora.pool.ntp.org iburst
server 1.fedora.pool.ntp.org iburst
server 2.fedora.pool.ntp.org iburst
server 3.fedora.pool.ntp.org iburst
# server xxx.yyy.www.zzz

And both ntpd and vdsmd started again and host up and SPM in webadmin.

Probably another thing to take care of for Fedora 18 configured as an
oVirt node.

Gianluca