[illumos-gate] 3409 mdb: Open source implementation of nfs module -- Vitaliy Gusev <gusev.vitaliy⊙gc>

whee:)

Thanks for your patience. Big upstream things like this benefit from documentation and stories of time-in-the-field.

Meanwhile it's time for lunch and my favorite workflow: "ssh $BUILDZONE pwait $PID_OF_BUILD ; say -a $EXTERNAL_SPEAKER Build is done, Dan"

:D

I have feeling, that mdb bit can be improved still;)

esp once 4.x will land

danmcd: "say" is a perfect tool, indeed.

tsoome: I expect you're right. And looking forward to 41... I had a branch of it in illumos-joyent, not long after nfs-zone went back. It likely needs to be massively rewhacked.

And yes @otis -> say(1) is one of MacOS's great features.

nice, didn't know about it

hrm. why.... ping: sendto Network is unreachable

reboot did fix it. weird.

Anybody know what ntpq means when it says a peer's refid is LOCAL(x)?

IIRC it's a local source only

so it could be the RTC

Even when the remote is an IP address?

And if a refid is LOCAL(x), does that mean there will be a subsequent line where the remote is LOCAL(x)?

it could be that the remote uses an undisiplined local source

e.g. i get it on my clients if both GPS sources and internet sources fail on my NTP server

it's only tracing locally as best it can from the last RTC read

aka very bad in my case

X is probably because it's not to be used unless no other source is available

Well the x is an integer

The line I'm parsing is this:

-216.218.254.202 LOCAL(1) 13151 5 u 646 1024 377 70.928 -4.351 1.224

That IP is clock.sjc.he.net, but when I use that as my time server, my refid isn't local.

the link also seems to indicate it's an undiciplined local source on the remote

ok the integer seems to be the X clock of that source type

I found this: serverfault.com/q/472455

gist.github.com/sjorge/f4b9f3e66a270198ab8557fd319aa5af e.g PPS(0) is my first local PPS source and GPS(0) my first GPS source, GPS(1) is my fallback GPS source, that one does not have a percise occilarion cristal

which doesn't exactly make things clear, but it makes me wonder if this user has a remote listed as LOCAL(x) to match that one with the refid of LOCAL(x)

Which, if they do and I don't fix it, I'm just moving the crash to the next line.

But I think in any case, it's possible for either remote or refid to be LOCAL(x), so I need my regexp to accommodate both.

the refid on a remote is what the it (remote) is using as it's primary source

with my above example from my server, my clients will show PPS(0) as the refid behind the remote ip

Ok, so then that should mean there won't necessarily be a line where the remote is LCOAL(x)

i think so yes

well i think you can have .LOCAL. LOCAL(0) ...

My regexp for that field is ([a-zA-Z0-9.]+)

which would be fine with the dots, but it's choking on the parens.

But I think to be on the safe side I should handle remote having parens in it as well.

see my gist from earlier

remote colum can certainly have () in it

Yeah.

OK, thanks, that's helpful.

sjorge: Can you tell me what's going on with line 7 in your gist?

@ptribble --> can't remember if you have push permission upon approval or not.

(Haven't RTI-reviewed your RTI, but knowing this a priori would be useful. )

you can look at people's roles in the bug tracker

OH!

Thank you Rich, sorry for the ping Peter.

I do (although being quiet haven't used it in a while)

bahamat: it's fqdn (ip) IIRC

but it usually gets truncated before it hits the ip

sjorge: Ok, so if you do -n for numeric only, then it shouldn't do that.

you might want to add -n i guess

Yeah, I do already

yeah, it also shows the PPS(0) and friends as the special ip ranges

I guess that should have been obvious earlier, but I was over thinking it.

Ok, so I may not actually have to handle it in the remote column.

But also, it doesn't seem like it would break anything if I do anyway.

so not ipv6 does very weer stuff too

*do note

Yeah, but I don't think we're doing anything with the remote and peer, we're just trying to get the delay/offset/jitter

i send you a privmsg with an ipv6 entry

as it's actually a global addr

i'm off to bed now though

Yeah...I should add :'s to my regexp so that those are supported