10:40:11 <solitario> Thank you for Java OpenJDK 17, 18 and 19 in OpenIndiana illumos. Very nice!
13:46:58 <jbk> anyone have an illumos box w/ a link aggregation setup that could check something really quick?
13:47:19 <andyf> I have one
13:47:55 <jbk> if you do dladm show-aggr -x
13:48:10 <jbk> does the speed for the aggr show as the speed of the individual links, or the sum of the links?
13:48:55 <andyf> It is not the sum - both of my links are the same speed so I don't know exactly what it is
13:48:57 <andyf> https://paste.ec/paste/eTph9tkB#THg7Ll2LoFabC3dHz0CeKKW3n65c0dvd3zAS+Oh5iU4
13:49:16 <andyf> Probably min(link speeds)
13:49:27 <jbk> ok.. thanks..
13:55:20 <jbk> i don't know if it was a local change that got lost in an upstream sync, or something that changed in illumos-gate
13:56:02 <jbk> ours (at least) used to show the sum of the links (so in your case the speed for aggr0 would be '2000Mb')
13:57:59 <jbk> though I think at least i've found the relevant portions of the aggr driver to dig through now
13:58:17 <andyf> My output was from OmniOS, unsurprisingly.
14:00:19 <andyf> It appears to come from the 'ifspeed' kstat
14:05:02 <jbk> yeah and it seems like the value comes from aggr_grp_t->lp_ifspeed
14:05:09 <dangergrrl> is there a reason that some of our standard network tools take -4 and -6 like linux and others reguire -A inet6 ?
14:05:26 <andyf> and I don't see any differences between SmartOS and gate in managing that value
14:05:28 <jbk> which based on the code is supposed to take the speed of the first link
14:21:52 <bahamat> jbk: On SmartOS it also only shows the single link speed.
15:35:07 <jbk> yeah.. and since we seem to change our minds every release how we use git (tags no branches no tags) I really can't get a good history on it.. but it looks like the 'first link speed' is the standard.. so that's a good enough answer
17:36:03 <jclulow> dangergrrl: I am sure some of it is an accident of history.  Some of it, though, is probably because there are more address families than just AF_INET and AF_INET6
17:38:17 <dangergrrl> jclulow, I am also guessing that Illumos/OI use some gnu tools and some legacy ones but it's slightly confusing when some tools require -A inet6 and others want -6
17:38:58 <jclulow> A distribution like OI might layer on some more stuff, but I don't believe we have any GNU stuff in the base OS
17:45:38 <dangergrrl> I still have things I want to fix on my Illumos bix but I don't have time for at least a week, I have deliverables I promised next Wednesday and I need the money :)
17:47:07 <dangergrrl> I have in fact reproduced similar ipv6 problems on my LAN on my raspi running linux 5.15 to the ones I'm having on Illumos so I suspect my ISP provided router is doing something wrong
19:19:13 <danmcd> Advertising new prefixes w/o deprecating the old ones, perhaps?
19:19:33 <danmcd> I'm glad we aren't misbehaving any worse than Linux.
19:20:32 <danmcd> As for -A inet and what not, Solaris was an early adopter of IPv6 (was in deep prototype before I arrived at Sun in 1996, and I'd worked on one of the two ancestral BSD IPv6s at NRL prior to Sun) so they picked what they thought was right at the time.
19:36:32 <dangergrrl> danmcd, new prefixes w/o deprecating might be the exact problem... On my raspi the old address doesn't show in ifconfig but it still breaks.  On Illumos both prefixes show up in ifconfig
20:12:02 <isoriano> Good evening.
20:17:48 <toasterson> Ola