On Thu, 16 Apr 2009, Alan Coopersmith wrote:
> Neal Pollack wrote:
> > On 04/15/09 11:46 PM, Dan Anderson wrote:
> >> Part of the problem (or "feature"), as you may know. is that NTP has
> >> some sanity checking code and it doesn't adjust for large time
> >> offsets. That has to be done manually (or automated with scripts)
> >> before the NTPD takes over. This is to prevent weirdness from rogue
> >> (or pirate) NTP servers.
> >>
> >
> > Which implies that the Solaris suspend/resume code/framework should take
> > this into account.
> > But it does not. For example, how hard would it be to add a "feature"
> > to the resume code path
> > that triggers something to the effect of;
> >
> > svcadm disable ntp
> > ntpdate "well-known-server"
> > svcadm enable ntp
>
> Wouldn't it be simpler to have ntpd listen for SIGTHAW, and when it gets
> it, just do a single resync-to-server-without-sanity-checks internally?
Maybe. I think that part of the problem is that the network
actually goes down for s/r, causing ntp to abort in some manner.
Possibly adding a SIGTHAW handler would mean that ntp waits for a
while for networking to properly resume before retrying. But even
there, what if there was no usable network after resume?
---- Randy
>
> --
> -Alan Coopersmith- alan.coopersmith at sun.com
> Sun Microsystems, Inc. - X Window System Engineering
>
> _______________________________________________
> laptop-discuss mailing list
> laptop-discuss at opensolaris.org
>
