[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: install/51214: Sysinst should add dhcpcd to rc.conf with -b flag.
The following reply was made to PR install/51214; it has been noted by GNATS.
Cc: install-manager%netbsd.org@localhost, gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost
Subject: Re: install/51214: Sysinst should add dhcpcd to rc.conf with -b flag.
Date: Sat, 4 Jun 2016 05:30:55 +0000
On Sat, Jun 04, 2016 at 04:45:01AM +0000, Robert Elz wrote:
> | Don't we have a notion of mountcritical which could work for NFS case?
> Those (mountcritlocal and mountcritremote) make sure that the mounts happen
> before they are required to be used elsewhere, they have nothing to do with
> delaying mount attempts until the network is ready.
I see, this situation is very tricky :-(
> | I'm unsure what is the reason for ntp/ntpdate to be so critical.
> Lots of things fail to work correctly if the time is not monotonically
> increasing. Unlike ntpd, ntpdate will step the clock, forward or backward,
> as needed (which is exactly what it should do, when run before anything else
> is started, or as little else as possible.)
I didn't know about that, sorry.
> NetBSD already boots to a runnable state much quicker than most
> other systems, and in general there are reasons for the sequence.
I do not have concrete numbers (and even if I did, we could argue about
their validity), but I feel like machines can and do complete a full
boot in the time it takes me to load nouveau or wait on ntpdate.
> | (while here, a third candidate is powerd).
> That ons potentially could start slower (I think) but does its startup time
> really matter? I cannot recall ever noticing any kind of delay from
> powerd starting up (there's probably more time spent writing to the console
> that it is starting, than actually starting it...)
I managed to find one example that's probably valid!! ;-)
Main Index |
Thread Index |