12:01:55 <gitomat> [illumos-gate] 15732 Cannot bring down zone due to lingering IP references -- Andy Fiddaman <illumos⊙fn>
17:10:24 <gitomat> [illumos-gate] 15452 bge MAC_PROP_MEDIA support -- Robert Mustacchi <rm⊙fo>
19:06:23 <gitomat> [illumos-gate] 15733 Windows client gets disconnected when copying files to SMB share -- Matt Barden <mbarden⊙rc>
19:23:18 <gitomat> [illumos-gate] 15720 pcidr: the comparison will always evaluate as 'true' -- Toomas Soome <tsoome⊙mc>
19:25:56 <bahamat> what's generally preferred, and why? routeadm, or just enabling smf services directly?
19:27:30 <richlowe> I usually use routeadm, and because it's there
19:28:42 <richlowe> I usually assume if there's a facility for the admin to do something, and another more manual way, the tool is going to be the safer, supported-er, path.
19:28:55 <bahamat> But, `routeadm -u -e ipv4-forwarding` is exactly the same as just `svcadm enable ipv4-forwarding`.
19:29:18 <richlowe> yup, I'm just not one to go off into the woods on a technicality.
19:30:07 <andyf> Does routeadm play with some SMF properties too? It's a while since I looked
19:31:08 <bahamat> Some things like that never sat well with me, but I highly prefer declarative configuration rather than imperative.
19:31:18 <bahamat> andyf: It can, but I don't need any of those.
19:31:35 <richlowe> the routeadm manual -- surprisingly to me -- says using svcadm too is fine
19:31:56 <bahamat> I'd rather just say that the service should be enabled rather than write a mini script to read the state and change it if necessary
19:32:08 <richlowe> it also says routeadm -u can't be used with other options though
19:32:13 <richlowe> so obviously there's some quirks in there :)
19:32:45 <richlowe> bahamat: I think that's what I did in ansible to kick routing on for tailscale, but I thought at the time "I should really have a routeadm module thingy for this"
19:35:30 <bahamat> I don't know, maybe routeadm is just syntatic sugar on top of svccfg for all the routing services. It's really hard to tell what the intended usage pattern is supposed to be.
19:35:42 <richlowe> I think that is basically it
19:35:51 <bahamat> But I also always hated SMF service properties. I like SMF a lot otherwise. I don't mind XML, but using svccfg to set/read properties...no thanks. Give me a flat file any day.
19:35:59 <richlowe> but I was surprised the service properties etc, are documented as also interfaces.
19:37:54 <bahamat> I remember using solaris 10, and the documentation for bind was to set all the smf service properties, then smf would modify your bind config for you. I *hated* that.
19:40:59 <nomad> And thus the benefit of not steeping in any given OS. I didn't even know routeadm existed until just now.
19:41:29 <richlowe> bahamat: the goal (then?) was that all that stuff could be in an SMF profile, you would apply.
19:41:32 <jbk> the flip side of that is that it handles persistent vs temporary, parsing, atomicity, etc for you
19:41:40 <jbk> which many, many things often screw up
19:41:43 <jbk> (even things in-gate)
19:41:46 <richlowe> which side-steps config management rather unfortunately.
19:41:46 <jbk> e.g. dladm
19:41:53 <bahamat> richlowe: Yeah, but it makes config management a nightmare.
21:36:00 <gitomat> [illumos-gate] 15721 rcm_daemon: the comparison will always evaluate as 'true' -- Toomas Soome <tsoome⊙mc>