On Wed, Mar 14, 2018 at 11:42 AM, Miroslav Lichvar <mlich...@redhat.com> wrote:
> On Wed, Mar 14, 2018 at 11:16:10AM +0100, Christian Ehrhardt wrote: > > On Wed, Mar 14, 2018 at 11:01 AM, Miroslav Lichvar <mlich...@redhat.com> > > wrote: > > > At this time, I'd be interested in including only in the first one. We > > > can reconsider the other two later if you are still interested. > > > > > > > Worst case we at least improve the messaging which is better than > nothing. > > > > That can be ok depending on what "later" means, what timeline are we > > talking about for "later"? > > In March is kind of ok, >=April would likely be too late for me. > > I think this all needs more discussion and I would like to postpone it > at least after 3.3, which will hopefully be released at the end of > March. > > To me it feels like there is a bigger problem that needs to be solved > first. Containers need more information about the system clock, which > only the host can provide. If this can be fixed (I'm not sure how), > maybe there will be a better solution for the problem that -X was > intended to fix. Looking form far away at the problem I think I agree. But all of this takes some (probably a lot) time. Let me finish V4 as a discussion example for later on and I expect only the improved output to be included for now. > > > The example unit file shouldn't change. > > > > > > > Well, without dropping ConditionCapability=CAP_SYS_TIME it will never > try > > to use it. > > No matter if one configures it for -x or the new -X - it will just not > even > > try while it would work inside the limits of -x/-X in those cases. > > The example unit file is intended for the typical use case, where > starting chronyd in containers (with or without -x/-X) makes no sense. > The thing that enables the -x/-X option should also remove the > ConditionCapability. If you will go with the wrapper approach, you can > modify the file in your downstream package. > Yeah, given the time constraints on "a real solution" I'll end up with my wrapper for now. And yes I can change things there as I needed. P.S. as just outlined on IRC I might even make -x (lower case) the default in containers. As otherwise container features will make it by default "run & fail" as the CAP will be around. Anyway - thanks for the discussion and I hope to get to the V4 soon. -- > Miroslav Lichvar > > -- > To unsubscribe email chrony-dev-requ...@chrony.tuxfamily.org with > "unsubscribe" in the subject. > For help email chrony-dev-requ...@chrony.tuxfamily.org with "help" in the > subject. > Trouble? Email listmas...@chrony.tuxfamily.org. > > -- Christian Ehrhardt Software Engineer, Ubuntu Server Canonical Ltd