How to make boot wait a few seconds more for device response?

[Ian Lepore]

On Sun, 2014-01-05 at 17:15 -0500, kpneal at pobox.com wrote:
> On Sun, Jan 05, 2014 at 02:16:19PM -0700, Ian Lepore wrote:
> > On Sun, 2014-01-05 at 21:06 +0000, Thomas Mueller wrote:
> > > > I don't know whether it's documented anywhere, I just remembered seeing
> > > > it in the source when I was working on a change to nfs root filesystems,
> > > > so I grepped and re-found it yesterday.  I think it's set in terms of
> > > > whole seconds.
> > > 
> > > -- Ian
> > > 
> > > This was for vfs.mountroot.timeout .  
> > > 
> > > There needs to be better documentation.  I have already searched for too many needles in the haystack and lost a lot of time that way.
> > > 
> > > I could type "show" at loader prompt, but figure only a small chance of finding everything, output may be sketchy.
> > > 
> > > So I still don't know whether it should be seconds or milliseconds.
> > > 
> > > Where did you find it in the source tree?  I made another search, unsuccessful, for the needle in the haystack.
> > > 
> > > Tom
> > 
> > Yes, for vfs.mountroot.timeout the units are whole seconds.
> > 
> > It's implemented in src/sys/kern/vfs_mountroot.c
> 
> Is this in 9.2-RELEASE? I'm looking at that file myself and I can't see
> where exactly that option is checked. I do see this:
> 
> 	parse_dir_ask_printenv("vfs.root.mountfrom");
> 	parse_dir_ask_printenv("vfs.root.mountfrom.options");
> 
> I also see ".timeout" checked when parsing input from somewhere. I can't
> tell where exactly it is coming from, though. But I can't see where it
> checks specifically for "vfs.mountroot.timeout".
> 
> Am I misreading it? Or is it not present in 9.2-RELEASE?
> 
> Sorry to be a pain on this, but I'm looking to avoid trouble with a 
> machine I'm upgrading remotely. I want to be sure before I pull the
> trigger.

Hrm, it looks like in 9.2-RELEASE it's just hard-coded to 3 seconds and
not tuneable.  In 9-stable it's tuneable.

-- Ian