On 11/25/2014 03:53 PM, Robert Moskowitz wrote:
On 11/25/2014 10:44 AM, Gordan Bobic wrote:
On 11/25/2014 03:28 PM, Robert Moskowitz wrote:
My cubieboards do not have a battery backing up the clock, so every boot
comes up with the date:
Jan 1 00:
But ntpd to the rescure to set the date. However, it seems the slew has
gotten too great:
Jan 1 00:01:33 medon ntpd[765]: 0.0.0.0 0617 07 panic_stop +154604256
s; set clock manually within 1000 s.
and now nptd is not running. I did set the date with 'date
MMddhhmmyyyy', but still need to restart ntpd.
It would seem to me that a hack around this is to run date in the boot
process. Something like:
date 010101012014
Just before starting of ntpd. But I know nothing of boot scripts, where
to put this command, and if that would be the proper format within the
boot process.
Can anyone lend me a hand on what/where to set this up?
I would have thought the correct way to deal with this would be to use
the ntpdate service in addition to the ntpd service. First ntpdate
sets the clock as a one off, then ntpd keeps it in sync periodically.
OK. What would I add to run ntpdate first?
yum install ntsysv
Then tick the ntpdate option and clock OK.
Or use chkconfig.
Thing is that the network
is not working when ntpd starts. I see in messages:
Jan 1 00:00:28 medon ntpd[765]: 0.0.0.0 c016 06 restart
Jan 1 00:00:28 medon ntpd[765]: 0.0.0.0 c012 02 freq_set kernel 0.802 PPM
Jan 1 00:00:28 medon ntpd[765]: Listen normally on 6 eth0
2607:f4b8:3:1:67:15ff:fe00:154 UDP 123
Jan 1 00:00:28 medon ntpd[765]: 2607:fcd0:daaa:daaa:daaa:daaa:daaa:daaa
interface fe80::67:15ff:fe00:154 -> 2607:f4b8:3:1:67:15ff:fe00:154
Jan 1 00:00:28 medon ntpd[765]: peers refreshed
Jan 1 00:00:28 medon nscd: 848 cannot stat() file `/etc/netgroup': No
such file or directory
Jan 1 00:00:28 medon NetworkManager[599]: <info> Policy set 'System
eth0' (eth0) as default for IPv4 routing and DNS.
Jan 1 00:00:28 medon nscd: 874 cannot stat() file `/etc/netgroup': No
such file or directory
Jan 1 00:00:28 medon NetworkManager[599]: <info> Policy set 'System
eth0' (eth0) as default for IPv6 routing and DNS.
Jan 1 00:00:28 medon NetworkManager[599]: <info> Activation (eth0)
Stage 5 of 5 (IP Configure Commit) complete.
Jan 1 00:00:31 medon ntpd[765]: 0.0.0.0 c515 05 clock_sync
Jan 1 00:01:14 medon dnsmasq[745]: reading /etc/resolv.conf
Jan 1 00:01:14 medon dnsmasq[745]: using nameserver
2607:f4b8:3:0:9254:5400:0:148#53
Jan 1 00:01:14 medon dnsmasq[745]: using nameserver 208.83.67.148#53
Jan 1 00:01:33 medon ntpd[765]: 0.0.0.0 0617 07 panic_stop +154604256
s; set clock manually within 1000 s.
I deleted the NetworkManager lines where the ethernet is coming up. ntpd
is ok with starting before the network is ready. Would ntpdate also wait?
No, it wouldn't. If your network isn't up before ntpdate starts you are
out of luck. The network service should be starting before ntpdate,
which should be starting before ntpd. What will disrupt this is reliance
on NetworkManager.
You need to configure yout network interface by creating a suitable
configuration file in /etc/sysconfig/network-scripts/ifcfg-<devicename>
Make sure you add the following options there:
NM_CONTROLLED=no
ONBOOT=yes
Then disable NetworkManager from starting up.
The only problem I foresee is that fsck may insist on running every
time because the last mount time will be in the future on every boot.
What would I look for in messages for this? I think I saw something
about this a time or two.
It'd come up on the console. fsck would run, then reboot, then it would
come up.
Do your systems have a battery backed up clock?
Yes, all of them.
Gordan
_______________________________________________
users mailing list
[email protected]
http://lists.redsleeve.org/mailman/listinfo/users
