SCIENTIFIC-LINUX-USERS Archives

February 2015

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Bill Maidment <[log in to unmask]>
Reply To:
Bill Maidment <[log in to unmask]>
Date:
Tue, 10 Feb 2015 18:44:33 +1100
Content-Type:
text/plain
Parts/Attachments:
text/plain (138 lines)
OK. The truth is out!
The power outage did not resolve the problem, even though the guests were shutdown and freshly booted.
The rtc on each SL7 guest on the SL7 host were set incorrectly as the localtime, because the XML file had:
<clock offset='localtime'/>

Changing the XML files for the SL7 guests to contain:
<clock offset='utc'/>
fixed the problem.

However, the SL6 guests on the SL7 host still needed:
<clock offset='localtime'/>
to get the time correct on them.

Some inconsistency here, but I can cope with that until the SL6 guests get replaced with SL7.

Cheers
Bill
  
 
-----Original message-----
> From:Bill Maidment <[log in to unmask]>
> Sent: Friday 6th February 2015 20:41
> To: Bill Maidment <[log in to unmask]>; [log in to unmask]
> Subject: RE: SL7 - crond starting before chronyd causing problem with mrtg {solved - I think}
> 
> 
> 
> Hmm
> 
> It seems that I forgot that kvm by default only suspends the guests when you reboot the host. This explains the delay in resyncronising the clock, etc.
> 
> I've now fixed up the /etc/sysconfig/libvirt-guests on the host to ensure a reboot of guests.
> 
> I'm now awaiting the next reboot (a scheduled power outage on Tuesday) to prove the point.
> 
> Sorry all for the noise; at age 68 the little grey cells aren't what they used to be.
> 
> Cheers
> 
> Bill
> 
> -----Original message-----
> From: Bill Maidment <[log in to unmask]>
> Sent: Thursday 22nd January 2015 18:26
> To: [log in to unmask]
> Subject: RE: SL7 - crond starting before chronyd causing problem with mrtg
> 
> Thanks David.
> However, even using After= chronyd.service the chronyd still takes too long to correct the time for the time-zone.
> 
> After some investigation, it seems that the KVM system is at fault when the host is rebooted, but not when the guest is rebooted.
> On reboot of the host KVM adds the timezone shift to the hardware clock for the guests and then the guest adds the timezone shift again.
> This extra addition of the timzone shift did not happen on SL6.
> 
> After running hwclock --systohc on the guest, I get the following setup.
> 
> [root@giggs2 ˜]# timedatectl 
>       Local time: Thu 2015-01-22 17:55:26 AEDT
>   Universal time: Thu 2015-01-22 06:55:26 UTC
>         RTC time: Thu 2015-01-22 06:55:26
>         Timezone: Australia/Sydney (AEDT, +1100)
>      NTP enabled: yes
> NTP synchronized: yes
>  RTC in local TZ: no
>       DST active: yes
>  Last DST change: DST began at
>                   Sun 2014-10-05 01:59:59 AEST
>                   Sun 2014-10-05 03:00:00 AEDT
>  Next DST change: DST ends (the clock jumps one hour backwards) at
>                   Sun 2015-04-05 02:59:59 AEDT
>                   Sun 2015-04-05 02:00:00 AEST
> [root@giggs2 ˜]#
> 
> On reboot of just the guest, I get the following, the RTC looks OK:
> 
> [root@giggs2 ˜]# timedatectl 
>       Local time: Thu 2015-01-22 18:05:09 AEDT
>   Universal time: Thu 2015-01-22 07:05:09 UTC
>         RTC time: Thu 2015-01-22 07:05:09
>         Timezone: Australia/Sydney (AEDT, +1100)
>      NTP enabled: yes
> NTP synchronized: yes
>  RTC in local TZ: no
>       DST active: yes
>  Last DST change: DST began at
>                   Sun 2014-10-05 01:59:59 AEST
>                   Sun 2014-10-05 03:00:00 AEDT
>  Next DST change: DST ends (the clock jumps one hour backwards) at
>                   Sun 2015-04-05 02:59:59 AEDT
>                   Sun 2015-04-05 02:00:00 AEST
> [root@giggs2 ˜]#
> 
> On reboot of the host, I get the following on the guest, the RTC looks wrong:
> 
> [root@giggs2 ˜]# timedatectl 
>       Local time: Fri 2015-01-23 05:15:41 AEDT
>   Universal time: Thu 2015-01-22 18:15:41 UTC
>         RTC time: Thu 2015-01-22 18:15:42
>         Timezone: Australia/Sydney (AEDT, +1100)
>      NTP enabled: yes
> NTP synchronized: yes
>  RTC in local TZ: no
>       DST active: yes
>  Last DST change: DST began at
>                   Sun 2014-10-05 01:59:59 AEST
>                   Sun 2014-10-05 03:00:00 AEDT
>  Next DST change: DST ends (the clock jumps one hour backwards) at
>                   Sun 2015-04-05 02:59:59 AEDT
>                   Sun 2015-04-05 02:00:00 AEST
> [root@giggs2 ˜]#
> 
> Regards
> Bill Maidment
> 
> -----Original message-----
> > From:David Sommerseth <[log in to unmask]>
> > Sent: Wednesday 21st January 2015 23:33
> > To: Bill Maidment <[log in to unmask]>; [log in to unmask] <[log in to unmask]>
> > Subject: Re: SL7 - crond starting before chronyd causing problem with mrtg
> > 
> > On 21/01/15 12:18, Bill Maidment wrote:
> > > Hi guys
> > > MRTG is having problems with timeshift because crond starts before chronyd on SL7.
> > > I have added chrond.service as a dependancy to the crond.service. Is that the right approach to solving the problem?
> > 
> > This sounds like the right approach.  But systemd have different levels of
> > dependencies.  I recommend you to check the dependency you used against man
> > systemd.unit.  Check out After=, Requires= and Requisite=.
> > 
> > --
> > kind regards,
> > 
> > David Sommerseth
> > 
> >
> 
> 

ATOM RSS1 RSS2