Re: [chrony-users] Two Issues

[Weary1 <chrony_01@xxxxxxxxxxxxxxxxxxxx>]

On 03/12/2015 06:04 AM, Miroslav Lichvar wrote:
> Is there a DNS server listed in /etc/resolv.conf before the network is
> configured?

Well, I can only look at it *after* the system is fully up & running.  But 
unless the system changes that file on-the-fly during the start-up process, 
the answer is "Yes -- two of them, actually."

However, the timestamp on that file is less than eight hours ago; so 
*something* touched it (it wasn't me).


> Older chrony versions (before 1.26) gave up too quickly if
> the name in chrony.conf couldn't be resolved.

That could indeed be part of the issue, at least on the server (which is 
running the default version for Debian Stable, 1.24-3.1+deb7u2).


> One workaround is to specify the servers with IP addresses instead of
> names.

I assume you mean the NTP servers defined in chrony.conf for chrony to use as 
its upstream sources.


> For the other system using 1.29, I'm not sure what could be the
> problem. In the "chronyc activity" output, is the number of sources
> with unknown address zero?

No mention of "unknown address" in response to a "chronyc activity" command; just:

    :~$ chronyc activity
    200 OK
    6 sources online
    0 sources offline
    0 sources doing burst (return to online)
    0 sources doing burst (return to offline)
    0 sources with unknown address
    :~$

HOWEVER...  This is *after* I restarted chrony, and it is running normally.  I 
can't check what it says when in its "confused" state until I re-boot that 
system -- and it happens to be the one I'm using at the moment to write this 
message, so that will have to wait.


> With recent versions, it should always work, even if it's started
> before the network is up and servers are specified by names. It's not
> necessary to use the offline option and switch it later to online via
> chronyc. The recommended configuration for most systems is:
>
> server 0.pool.ntp.org iburst
> server 1.pool.ntp.org iburst
> server 2.pool.ntp.org iburst
> server 3.pool.ntp.org iburst
> driftfile /var/lib/chrony/drift
> makestep 10 3
> rtcsync

I'm using different NTP servers, which I doubt matters.  But they are NOT 
tagged with the "iburst" directive.  Would adding that make a significant 
difference to my issues?

Also, probably because these are Debian (or Debian-based) installations, the 
driftfile path is:

    driftfile /var/lib/chrony/chrony.drift

(There is a comment in the default chrony.conf, presumably from the package 
maintainer, to the effect that this was done to comply with Debian filesystem 
standards.)

The "makestep" and "rtcsync" commands do not appear at all.

(BTW...  All the above comments are with respect to the chrony.conf on the 
server, with the older version of chrony, as that's easier for me to "get to" 
at the moment.)

Note that I've deliberately made only the minimum changes to the default 
chrony.conf that (I thought) I needed for a particular purpose.  Hence I did 
NOT remove those directives, as I had no reason to.  They were not there by 
default.  Are these discrepancies the result of the default chrony.conf having 
"evolved" over time, or has the Debian maintainer made more extensive changes 
than I'd presumed.


> The bigger hammer is the chronyc makestep command or the makestep
> directive in chrony.conf to step the clock automatically.

So which is the *better* (if perhaps not "bigger") hammer:  "makestep" or 
"initstepslew"?  IOW, which of them should I put into chrony.conf?

Thanks for your help.


--

Weary1


--
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx 
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx 
with "help" in the subject.
Trouble?  Email listmaster@xxxxxxxxxxxxxxxxxxxx.