00:48:08 [illumos-gate] 17473 Update JEDEC vendors to JEP106BM -- Robert Mustacchi 00:48:08 [illumos-gate] 17474 June 2025 hwdata update -- Robert Mustacchi 08:56:12 erm, how do I initialize my iscsi initiator iqn? iscsiadm list initiator-node shows no value for "Initiator node name" 08:56:51 and since I use host groups for the views of iscsi luns, I need the iqn of this client... 10:30:34 I think our iscsi stack is not super solid...on the target side, I have one target with 65534 sessions (probably max) and I can't offline that target, I can't even stop the target service because the stop method hangs 10:31:52 and according to https://src.illumos.org/source/xref/illumos-gate/usr/src/cmd/iscsitsvc/iscsitsvc.c?r=607aa931#170 it just opens the device and does an ioctl on it 10:33:34 which brings me to the conclusion that a reboot will be the only solution...correct? 10:37:35 probably... and yes iscsiadm list initiator-node should show the iqn, but it does not. 10:53:56 tsoome_: oh, same for you? is there another way to get the initiator iqn? 11:16:31 I did pick it from target side, it did appear on unlimited lun or something 11:58:17 tsoome_: hehe, ok 13:48:30 ok, we do get nice L"iqn.1986-03.com.sun:01:583fa0d80bfe.6769fa50" from ioctl(), we use vswprintf() in libSwprintf() to store this and end up having L"\x2e6e7169\x36383931\x2e33302d\x2e6d6f63\x3a6e7573\x353a3130\x61663338\x30386430\x2e656662\x39363736\x30356166" and fprintf("%ws") prints basically garbage (11 times byte 0 according to hexdump -C). 13:49:09 oh, how sweet... 13:50:36 either vswprintf or fprintf does something bad; fprintf("%ws") does print L"iqn.1986-03.com.sun:01:583fa0d80bfe.6769fa50" as expected, however. 13:52:05 more serious is the issue that I have 65534 sessions on this target - probably because the initiators do not log out correctly until all sessions are in use... 13:53:03 and of course then the target service is not even able to offline that target anymore 13:53:35 I do not know the details about iscsi protocol, but I'm pretty sure the sessions should be possible to get reaped... 13:54:00 I think also the target side should reap those 13:58:32 yep, in networked environment you can not assume every client system will end the session "correctly". 14:09:35 it seems, in iscsiadm, the fprintf("%ws") is to be blamed, wprintf(L"%s") does print the iqn correctly. 14:20:07 tsoome_: shall I open the bug and send a patch, or do you want to do it yourself? 14:21:21 please do - feel free to use my findings:) also it may be Woodstock can tell thing or two about it 14:21:36 ok, I'll take notes 14:22:05 I'm in a bit of in middle of other issue:D 14:23:07 gotcha ;) 14:23:33 I will also not file it before monday I guess...I need to file travel claims 15:25:21 got illumos installed on a small Alder Lake intel system that I'm bringing in as a router. some console quirks (serial console wasn't quite functional during install) but seems to be working. 15:28:36 (protectli VP6650 with coreboot firmware with an intel i5-1235U cpu). connected to a newly installed 10G fiber service (sonic.com) that.. isn't working with a 10G link speed but seems to be ok at 2.5G 15:36:26 NIC ? 15:36:52 Alder lake small means it's probably a I225 or I226 (aka. igc(4D)). 16:28:25 danmcd: 4x I226-V (copper), 2x X710 10Gb SFP+ 16:35:44 this cpu has 2 hyperthreaded "performance" cores and 8 single-thread "efficiency" cores. Seems to be behaving itself with all 12 threads enabled. 16:36:03 (but haven't really pushed it) 16:36:31 did fall back from 10 to 2.5, skipped 5G? 16:38:16 no. doesn't work when plugged into a 10G interface (with a different machine it eventually got a dhcp lease but had 85% to 95% packet loss) 16:39:04 when plugged into a 1G port on the other machine it works fine at 1G. when plugged into a 2.5G igc port on this machine it also works fine. 16:39:47 I see... 16:41:16 the ONT Sonic installed looks to be an Adtran 632V 16:55:54 (and unless I'm missing something it doesn't look like I can force the X710 into 2.5G or 5G mode - might be a limitation of the particular 10GbaseT sfp I have) 17:06:11 might be, yep. 17:10:33 okay, this is weird. connected it to a switch that is 1/2.5/5/10 capable and set the switch advertise only 2.5G. switch thinks the link came up at 2.5; illumos on the other end thinks it's 10G 17:16:31 iperf agrees with the switch.. 17:17:22 that may hint why 10G has problems? 17:18:02 no, it seems to work OK when 10G is enabled on the switch 17:40:04 You're plugging in one of the X710s to the switch, I'm assuming? 17:40:54 Or the I226/igc? If the igc, I think we only support 2.5G on that. If rmustacc is here and has cycles he may have more useful data in his L1-3 caches. 17:41:42 Correction: I226 ONLY supports up to 2.5Gbit. 17:41:44 I'm trying a bunch of different combinations to figure out what works and doesn't work before I call sonic. 17:42:13 ideal config is one X710 connected to sonic, the other to a switch. 17:42:28 current config is one X710 to switch, one igc to sonic at 2.5G 17:42:53 What does `prtconf -v | grep X7` say? Model & PCIe ID would be useful. I know there's some Intel common code that needs updating in i40e(4D) at some point. (It might actually fix fenix illumos#13230 ?!?) 17:42:55 code review '13230' not found... 17:43:02 I want to test one X710 to switch, other X710 at 2.5G to sonic just to fully cover the bases 17:43:19 value='Ethernet Controller X710 for 10GbE SFP+' 17:43:19 value='Ethernet Converged Network Adapter X710' 17:43:19 value='Ethernet Controller X710 for 10GbE SFP+' 17:43:20 value='Ethernet Converged Network Adapter X710' 17:43:45 Q from the peanut gallery: aside from a blue hedgehog that drops coins everyhwhere, what is sonic? 17:44:21 sonic.com. ISP that is offering 10G residential fiber 17:44:28 a drive-in fast food chain? :) 17:44:33 that too 17:44:48 oh man, I haven't had a sonic chicken fried steak in decades. 17:44:55 Kind of miss that badness. 17:45:05 sommerfeld, ok, I am jelous. 17:45:20 * nomad was happy with his 1G fiber but nope, not any more. 17:45:33 the co-op only goes to 5gb right now, but i wouldn't be surprised if they up it in the future 17:45:47 though I just have 1gb right now because i haven't found a need for anything faster (yet) 17:47:46 Ooops, wrong command: `prtconf -d | grep X7` ==> my bad. 17:48:03 pci8086,0 (pciex8086,1572) [Intel Corporation Ethernet Controller X710 for 10GbE SFP+], instance #0 17:48:03 pci8086,0 (pciex8086,1572) [Intel Corporation Ethernet Controller X710 for 10GbE SFP+], instance #1 17:48:10 * danmcd is wondering if VZ FiOS can bump to >= Gig yet... 17:48:17 Okay, thanks. 17:48:52 jbk, I don't even need 1G but I *want* it. :) 17:49:20 and it's less than half the cost of what i was paying for comcast at my apartment too 17:49:56 granted no tv service, but i was down to like 2-3 channels at most, and I can always do OTA for stuff I don't want to stream 17:50:50 My options are Century Link or comcrap. I refuse to do business with comcrap so here I am. 1G fiber is the fastest they offer. 17:50:56 at least I have a static IP. 17:54:54 i got lucky -- the front half of the neighborhood is serviced by a company with the suitably generic 80s conglomerate sounding 'Consolidated Communications' which apparently is fairly terrible 17:55:21 but the back half is in another county, so it's power is from a local co-op, who also provides fiber internet to their service area 17:55:48 (I could still get consolidated as well if i wanted to, but given their reputation, unless I like really wanted to have redundant links, not sure it's worth it) 17:56:45 unfortunately, they don't do IPv6 yet is i guess one downside 17:57:18 but no cgnat, so.... 17:58:00 anyhow I can't seem to force the X710 link to 2.5G or 5G from the illumos end. 17:59:48 jbk, I'm using he.net for an IPv6 tunnel. It works fine as long as my IPv4 connection is stable. pfSense does a good job of connecting to it. 20:18:50 Is there a protocol on dev-guide submissions? 20:27:15 used to just be a PR 20:28:53 There IS a PR. One reviewer, but unsure if we needed more or not. 20:29:11 oh, if you want two for completeness gimme a link 20:29:27 https://github.com/illumos/dev-guide/pull/54 And the reviewer isn't +1-ing yet but I did fix his one concern. 20:35:08 commented