but I guess if it went out previously it's probably good I didn't resend it to the same list. That would have been rude of me.

you probably want the developer list to get attention

I am so very much not a dev (which you should all be thankful given my lately demonstrated attention to detail) but if I don't get a response to this post I suppose I should go there.

I hate being such a bother.

nomad: Gordon Ross might know where to start looking for this.

nomad: problem is that there are so many layers involved. if you could better identify *what* request from the servers is triggering the complaint from the AD server you might have a better chance of figuring out where this is coming from in our stack.

[illumos-gate] 17853 the last vnic will not go gentle into that good night -- Ryan Zezeski <ryan⊙zc>

i can't be the only one who thinks that that subject line is cute?

I got the reference :)

i actually don't

but i kinda think it's cute

[illumos-gate] 17852 want better debugging for asy(4D) -- Dan Cross <cross⊙oc>

SarahMalik: en.wikipedia.org/wiki/Do_not_go_gentle_into_that_good_night

anyone know offhand if an event (or maybe something monitorable via contracts) happens when svc.startd fails to stop a service (times out)?

I think, FMA can get it.

but it may have been on oracle solaris where I did poke it...

it might be nice to be able to automatically get a stack trace of processes in the contract if they're not stopping

(sorry about the delay.. got interrupted mid-thought :P)

like write them out to the service log or such...

it would be nice to have probe method (yea, I have spent too much time with solaris clusters;)

could modify svc.startd to do it, but I could also see where that might be going a bit out of scope as well.. just wondering of a nice way to do it...

could you write a dummy service that doesn't die, and ctwatch for it?

that should show you anything contract-y right?

jbk: illumos.org/issues/17059 might be interesting?

→ FEATURE 17059: libscf should offer a committed mechanism to poll for changes (New)

hmm.. maybe

it would be nicer to have a service that could just see a service is 'stuck' and then run pstack (or maybe even fancy an mdb -e "0t$PID::pid2proc | ::walk thread | ::findstack -v"