03:06:42 I have a trashtop that I would like to run a flavor of unix on. I see there are multiple forms of illumos, which should I use? It will be for mainly genral use stuff, not a daily driver, and seeing if I can get my code to run on it. (also possibily some packaging if it is avaible) 04:25:23 Dotz0cat: illumos has "distros" similar to how Linux does. If you're running a desktop system you might find OpenIndiana suitable. If you're running a server, probably OmniOS, and if you're doing mainly virtualization, then SmartOS. 14:37:20 I'm getting quite worried about the dead silence on https://www.illumos.org/issues/15302 - Our server vendor says the older cards are getting hard to get. They're talking boneyard at this point. 14:37:21 → BUG 15302: broadcom 3108 kernel driver update (New) 14:37:36 Is there anything that can be done to push this along? 14:47:10 i'm currently working on a lmrc driver for the latest generations of megaraid controllers, but that probably won't help you with 3108 14:48:45 the part I don't understand about illumos #15302 is that it says the system is unstable, but there's no description of how that instability manifests 14:49:11 given that mr_sas complains about unsupported features in the HBA, I'd expect it to not attach at all 15:06:14 It boots, it runs, but it accumulates errors in the HBA's internal log. 15:06:40 eventually, when the host is rebooted (e.g. for patching) it refuses to boot until those errors are cleared. 15:06:56 I'll add that to the ticket. 15:07:18 interesting 15:08:17 I think I have some screen shots from when I was working with SuperMicro to try to fix the problem. If I find them I'll add them to the ticket. 15:19:08 let me know if there's anything else that's unclear, please. 15:21:57 (I'm trying not to clutter the ticket with all the side discussion so may have trimmed too much.) 15:23:04 e.g. I have screen shots of the entire process of clearing the error if you want to see them. I somehow doubt that is the case, however. 16:28:10 [illumos-gate] 15265 loader: add mechanism to set os console based on last input device -- Toomas Soome 18:43:31 [illumos-gate] 15313 document pcitool register peek/poke mode -- Hans Rosenfeld 18:46:26 [illumos-gate] 14956 nvmeadm list improvements for namespace management -- Hans Rosenfeld 19:12:26 nomad: about issue with hba 3108 - did you try different firware IT for it ? 19:12:53 igork, if you mean "the latest from the vendor" then yes. 19:13:15 Most 3108s don't support IT firmware. 19:13:36 nomad: nope, i mean try to older different version and test it. also ticket missed firmware version 19:14:17 igork, I'll add the firmware version to the ticket. 19:14:34 We didn't try older firmware as that's not a good long term solution to the problem. 19:15:01 yeah, that doesn't like a solution at all really 19:15:27 nomad and list of others versions what you tested. if will be interest if your issue can be fixed by older firmware 19:23:31 The card is available to the kernel developer who wants to work on this problem. They're welcome to test that but, again, that isn't a good long term solution. 19:25:30 Without docs, it's pretty hard to help. 19:26:25 what's doing the SCSI<->SAS translation? as i noted in the ticket, the error seems to be related to that 19:27:15 err SATA 19:27:19 (monday :P) 19:27:51 is there an expander in there that might have firmware that needs to be updated? 19:30:16 ticket has been updated with firmware information. 19:33:00 jbk, sorry, didn't see your update until just now. If there's a way to ask prtconf for that information I'm happy to do it. The only thing we replaced was the HBA itself so anything else involved is still in there. 19:34:20 I'm pretty sure prtconf doesn't know, something else might. 19:35:02 everything in that particular host is SAS. It's a SSD-only fileserver. 19:35:41 there's sas/ses/etc knowledge in the FM topo, but I'm not sure if it includes things like this I'm afraid. 19:35:57 I'm willing to fish if someone hands me the rod and line. 19:36:27 It has been a very, very long time since I've been proficient with Solaris so I'll need direction from those who remember how it's done. 19:36:38 /usr/lib/fm/fmd/fmtopo will list the FM topology, and -V will make it verbose, but I'm not sure it'll tell you anything useful. 19:36:53 it's just more likely to know than prtconf. 19:37:11 that's a lot of output. 19:39:59 Who maks the system? 19:40:03 *makes 19:40:09 SuperMicro 19:40:12 What model? 19:40:37 (This is to try and answer your expander question) 19:40:48 should I paste the output of fmtopo -V to a pastebin? 19:41:28 Let me go find the quote to see what they call it. 19:42:42 SSG-6049P-E1CR36H 8600694390 / SERVER 19:42:58 and that matches the top of the tmtopo output. 19:43:07 hc://:product-id=SSG-6049P-E1CR36H:server-id=hvfs1:chassis-id=S287680X2712897/chassis=0 19:43:08 group: protocol version: 1 stability: Private/Private 19:43:08 resource fmri hc://:product-id=SSG-6049P-E1CR36H:server-id=hvfs1:chassis-id=S287680X2712897/chassis=0 19:43:08 group: authority version: 1 stability: Private/Private 19:43:08 product-id string SSG-6049P-E1CR36H 19:43:09 chassis-id string S287680X2712897 19:43:11 server-id string hvfs1 19:43:13 group: system version: 1 stability: Private/Private 19:43:15 isa string i386 19:43:17 machine string i86pc 19:44:32 smbios -t3 will also report that as well in the future. 19:44:41 But yes, I can tell you that server uses SAS expanders. 19:44:52 I used to use it a lot in a past life. 19:45:17 Simliarly, you should see some things in topo there, IIRC. Though the sestopo command would confirm it as well. 19:45:36 Especially if you only have a single 3108 in the box. 19:45:47 is this the expander by any chance? It's listed between the 3008 (the current HBA) and what appears to be the first SSD. 19:45:50 hc://:product-id=SSG-6049P-E1CR36H:server-id=hvfs1:chassis-id=S287680X2712897/motherboard=0/hostbridge=4/pciexrc=4/pciexbus=59/pciexdev=0/pciexfn=0/iport=0 19:45:50 group: protocol version: 1 stability: Private/Private 19:45:50 resource fmri hc://:product-id=SSG-6049P-E1CR36H:server-id=hvfs1:chassis-id=S287680X2712897/motherboard=0/hostbridge=4/pciexrc=4/pciexbus=59/pciexdev=0/pciexfn=0/iport=0 19:45:50 group: authority version: 1 stability: Private/Private 19:45:51 product-id string SSG-6049P-E1CR36H 19:45:53 chassis-id string S287680X2712897 19:45:55 server-id string hvfs1 19:45:57 group: storage version: 1 stability: Private/Private 19:45:59 initiator-port string w5003048023d621f1 19:46:01 initiator-port-pm string ff 19:46:15 right now it has a 3008 since we needed it to work and it was failing with the 3108. 19:51:39 The iports aren't expanders. 19:51:56 Those are the phys on the 3108. 19:52:59 this host doesn't have a 3108 in it right now. 19:53:25 Oh, well, then it'll be 3008s PHys. 19:53:31 * nomad nods 19:54:00 That box with a 3008 was what we used for a while at Joyent, which is why it has some additional topo bits there like drive bay location info, etc. 19:54:48 would sas3ircu give the expander info? 19:55:20 I think if you just want expander info can you gist/pastebin fwflash -l? 19:56:09 Should I just put that in the ticket? 19:56:13 Wherever. 19:56:49 https://pastebin.com/yzFZz85F 19:57:37 Those first two devices are expanders, IIRC. 19:57:44 thanks 19:57:56 I'll add that to the ticket. 19:58:12 Most expanders support SES (scsi enclosure services) which is how we enumerate and can update them. 20:02:20 Are the expanders build into the drive backplane, perchance? 20:08:14 Yes. 20:41:55 those error messages in the ticket are from the HBA log, correct? The CDB in the message is an ATA-PASSTHROUGH(16) CDB... so not sure why it'd be trying that if it's not a SATA disk 20:42:27 jbk, they came from the HBA, yes. 20:44:06 that host has 2 INTEL SSDSC2KB24 and 7 Micron_5300_MTFD and that's it. 20:45:04 IIRC, the Intel SSDs are directly connected to the motherboard. They're for rpool, everything else is in ssdp0 pool. 20:45:10 and the sense key appears to be basically a pass-through result as well.. 20:45:51 nothing in the OS will send that CDB 20:45:58 do you have smartmontools installed? 20:46:16 evidently not. 20:46:48 that and some of the sg3 utilities are the only things i'm aware of that'd issue that sort of CDB 20:47:09 oh wait, those smartmontools. Yes, I do have that installed. 20:47:13 * nomad was thinking of something else. 20:47:26 442 ? S 0:00 /opt/ooce/sbin/smartd --pidfile=/var/opt/ooce/smartmonto 20:47:36 I asked this yesterday but I did not get any reply that I saw. I have a trashtop that I would like to run a flavor of unix on. I see there are multiple forms of illumos, which should I use? It will be for mainly genral use stuff, not a daily driver, and seeing if I can get my code to run on it. (also possibily some packaging if it is avaible) 20:47:48 could be worth trying to turn that off to see if the errors go away 20:48:04 jbk, the card is no longer in that host. 20:48:11 oh 20:48:18 though in a month or so I'll be retiring an older file server and can, in theory, install the card there. 20:48:36 I'd prefer to hand the card over to someone who can update the kernel driver to match the new firmware features, though. 20:48:43 i wonder if it's maybe misidentifying the type of drive and issuing those CDBs 20:48:57 and the HBA can't handle the response 20:48:58 jbk, we needed the host to be functional so bought a 3008 and swapped the 3108 out. 20:49:38 SuperMicro said it was this warning in the card's log: 35992: 22-12-19,18:46:59 WARNING:Host driver needs to be upgraded to enable extended LD support 20:49:50 (on illumos, both SCSI and SATA drives 'appear' like scsi disks, and the SATA framework translates SCSI commands to SATA) 20:50:25 so smartd has to look at other bits to determine if it needs to send an ATA PASSTHROUGH command 20:50:51 (though our support for ATA-PASSTHROUGH(16) is also currently busted for most things) 20:58:56 Dotz0cat: If you want a basline system then probably illumos. The other Solaris options are generally based on that. (I speak in very broad swaths here). As for "which Unix", that's a much bigger question. 20:59:26 I'm not even up on the complete list of Illumos-based distros, let along all the Unix options. 20:59:36 I run FreeBSD and I'm happy with it. Take from that what you will. 21:00:08 I guess, what is a trashtop and do you care about a graphical console? 21:02:01 Most of the time when I hear people talking about a trashtop it's just old'garbage' laying around. E.g. 10+year old laptop they had in storage. 21:05:07 Generally I would say that making use of older hardware is more Tribblix territory than the other distros 21:06:02 Is there something that you've done differently on x86? 21:06:49 10 years ago at this point is inbetween Ivy Bridge and Haswell. 21:08:28 Yes a trashtop is a old garbage laptop. It does endup being x86_64 21:09:52 Although I would regard laptops as being generally not an ideal target for illumos 21:12:32 or desktops in general;) 21:17:36 It's not specifically age, but Tribblix tends to handle resource-constrained (especially small memory) systems fairly well 21:19:08 Although looking around my lab, the fact that my current SPARC box is one of the most modern devices might be a bit worrying 21:21:15 Ill look into tribblix. Part of the fun is getting stuff that shouldn't work to work 22:33:16 [illumos-gate] 15476 csh should be mediated in the packaging -- Dominik Hassler