andyf, tsoome: With my apologies, another small addendum to code.illumos.org/c/illumos-gate/+/2355/8

→ CODE REVIEW 2355: 14978 ZFS autoexpand property should work for root pools (NEW) | illumos.org/issues/14978

jclulow np:)

OmniOS stable r151044 has been released - omnios.org/rn/r44

sjorge: all of this leave/join by neuroserve made me miss it until an archive check. THanks.

Can @rmustacc and @jbk take another peek at code.illumos.org/c/illumos-gate/+/2487 ? I put in goodies for usr/src/test to go along with it.

→ CODE REVIEW 2487: 15146 sadb_unlinkassoc() isn't idempotent (NEW) | illumos.org/issues/15146

danmcd: Yeah, will do.

Thanks for adding tests.

[illumos-gate] 15100 WPTS BVT_SMB2Basic_ChangeNotify_... (write cases) -- Gordon Ross <gwr⊙rc>

[illumos-gate] 15067 linker set should be declared WEAK -- Toomas Soome <tsoome⊙mc>

danmcd: I'll try... but i'm getting beyond swamped at the moment

(of course the days right before i go on vacation)

(of course!) No worries. It's mostly about adding a test to usr/src/test.

this one is especially bizarre... i'm getting more convinced some sort of endpoint software is screwing things up, but can't prove it conclusively...

Ah that lovely burden of proof!

actually you might happen to know this offhand (i'll be source spelunking if not)... is there a dtrace probe that fires when a TCP connection is closed for any reason?

Maybe this?

14905 tcp ip tcp_closei_local state-change

Otherwise use FBT for something like tcp_close_common or something.

(hence source-diving)

hmm that might work

hrm.. an sdt probe for icmp messages would be nice.. we have one when we bump the mib counter

but to have the IP info...

oh this is annoying (but maybe already known to others).. we define ICMP stuff in netinet/ip_icmp.h and inet/ip.h in very similar, but not exactly teh same way

andyf: hey, I think you know this. How do the SMF seeds become repository.db?

danmcd: as far as you're aware, any inbound packet dropped (assume no IPsec policies or ipf in the picture) goes through ip_drop_input, right?

don't we have dtrace probes for drops?

they're implemented there

but appear to be somewhat useless due to #13578

but instead of the sdt probe there, can instead just use the slightly more involved fbt probe + decode fn args

richlowe: usually by a simple cp command unless you only have the xml files then you need something like manifest import

I need more detail than that, unfortunately

(I know about the copying, though)

In what context?

without the copying, I had it working -- which I'm told is surprising and/or mystical.

richlowe: checkout github.com/illumos/image-builder/blob/main/src/main.rs#L1523-L1584

then it broke for reasons that aren't worth going into, and those mysteries became important.

I'm not yet following

I filled andy in elsewhere, and unfortunately, we're just both confused

I am assuming bitness and nvlist binary format magic issues?

deeper than that, sadly

but the fact I was previously pulling off something it seems everyone believes to be magic is making it harder to debug.

because everyone just tells me oh, that doesn't work

and it did! it did!

You're a wizard Richard

computers can sense when you are getting a little too happy, that's when they decide to take you down a notch or two

./computers/girlfriends/ lawl

yeesh

j/k :)

<3

oops pardon latency jbk, yes it should go through ip_drop_input.

ok