http://bugzilla.suse.com/show_bug.cgi?id=1145193 http://bugzilla.suse.com/show_bug.cgi?id=1145193#c60 --- Comment #60 from ravas mi <ravas@tuta.io> --- (In reply to Reinhard Max from comment #59)

(In reply to ravas mi from comment #57)

...

I don't understand the point of "waitsync" on boot. Why do we need slewing at the first clock update?

"waitsync" has nothing to do with the question of stepping or slewing, it only makes chronyc block until either

a) chronyd has synchronized the time (regardless of the method it uses to get there)

or

b) the timeout has been reached.

"waitsync" seemingly only uses slewing. I meant that "makestep" seems like an alternative to chrony-wait/waitsync/slewing.

...

Why not use "makestep" to immediately make the correction?

We already have that in chrony.conf by default (as you already noticed yourself in comment 51), but if there is no network during boot chronyd cannot synchronize the time, even if you allow it to step.

It might indeed be worthwile to think about moving from "makestep" to "initstepslew", but that's irrelevant for the "no network at boot" issue we're discussing here.

It seems relevant. If either "makestep" or "initstepslew" fail to connect to an NTP server, then the boot continues. Why can't "makestep 0.1 1" replace the chrony-wait service? In testing, if I use "initstepslew" with a URL, then it always produces "Could not resolve address of initstepslew server <URL>." The documentation's example uses URLs, and if I use the IP of the server instead, it works -- implying the DNS isn't working at that point. If we could solve that, then "initstepslew" seems like it could also replace chrony-wait. -- You are receiving this mail because: You are on the CC list for the bug.