September 2, 2013 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | >
September 2, 2013 [00:04:38] *** mamash has left #smartos[00:07:26] *** szaydel has quit IRC[00:08:32] *** tbatchelli_m has quit IRC[00:08:42] *** szaydel has joined #smartos[00:21:54] *** tbatchelli_m has joined #smartos[00:30:06] *** elijah-mbp has joined #smartos[00:36:28] *** potatosalad has joined #smartos[00:46:58] *** tbatchelli_m has quit IRC[00:52:31] <konobi> my intel nuc seems to be doing pretty well so far[00:55:53] <richlowe> konobi: oh sure, you say that in here[00:56:03] <konobi> o.O[01:12:06] *** nahamu has quit IRC[01:12:50] *** nahamu has joined #smartos[01:22:16] *** szaydel has quit IRC[01:31:45] *** szaydel has joined #smartos[01:35:13] *** cjones_ has joined #smartos[01:48:37] *** potatosalad has quit IRC[01:55:35] *** groundwater has quit IRC[02:18:36] *** potatosalad has joined #smartos[02:20:49] *** potatosalad has quit IRC[02:21:51] *** potatosalad has joined #smartos[02:30:45] *** cjones_ has quit IRC[02:58:07] *** sa5bke has quit IRC[02:59:45] *** sa5bke has joined #smartos[03:05:23] *** potatosalad has quit IRC[03:07:44] *** potatosalad has joined #smartos[03:22:32] *** cjones_ has joined #smartos[04:12:10] *** szaydel has quit IRC[04:40:52] *** cjones_ has quit IRC[04:40:58] *** cjones__ has joined #smartos[04:54:22] *** szaydel has joined #smartos[04:55:20] *** Fuzai_ is now known as Fuzai[05:01:53] *** leecallen35 has joined #smartos[05:06:12] *** leecallen has quit IRC[05:13:37] *** zr0 has joined #smartos[05:13:48] *** kollross has quit IRC[05:16:11] *** kollross has joined #smartos[05:22:00] *** potatosalad has quit IRC[05:31:03] *** benbangert has quit IRC[06:15:29] *** axonpoet has joined #smartos[06:31:34] *** szaydel has quit IRC[06:36:05] *** patrickmslattery has joined #smartos[06:37:41] <patrickmslattery> Hi, wondering if anyone can give me pointers on where I am going wrong with a multi NIC setup[06:38:59] <patrickmslattery> I'm trying to setup a SmartOS zone so that it has a connection to the external world and to a private network that has an AD server setup (for JIRA user auth testing)[06:39:35] <patrickmslattery> I can't seem to get the private NIC attached no matter what I try[06:40:15] <rmustacc> So you have two different physical nics in the host/global zone?[06:40:25] <rmustacc> Or at least two, one for each of these types of traffic?[06:41:14] <rmustacc> I just want to make sure I understand what this looks like in the host before we go into the zone configuratin.[06:41:16] <patrickmslattery> Yes, first NIC is external, second is to the AD subnet, these are both VMs in a VirtualBox environment[06:41:17] <rmustacc> *configuration[06:41:47] <patrickmslattery> [root@08-00-27-a3-a2-ed ~]# dladm show-phys[06:41:47] <patrickmslattery> LINK MEDIA STATE SPEED DUPLEX DEVICE[06:41:47] <patrickmslattery> e1000g0 Ethernet up 1000 full e1000g0[06:41:47] <patrickmslattery> e1000g1 Ethernet up 1000 full e1000g1[06:41:56] <patrickmslattery> g0 is to AD[06:42:01] <patrickmslattery> g1 to external[06:42:17] <patrickmslattery> I added this to /usbkey/config[06:42:18] <rmustacc> Sure, okay. And do you have two different nic tags based on that?[06:42:25] <patrickmslattery> ad_nic=8:0:27:a:5:63[06:42:34] <rmustacc> Can you use pastebin/gist/something else rather than just dump it all?[06:42:50] <patrickmslattery> Sure, firing up gist[06:43:15] <rmustacc> Include nictagadm list and dladm show-phys -m if you can.[06:45:11] <patrickmslattery> https://gist.github.com/patrickmslatteryvt/6409264[06:45:52] <rmustacc> Okay.[06:46:00] <rmustacc> That all looks good.[06:46:26] <rmustacc> And the zone configuration you're tyring to use?[06:46:55] <rmustacc> That's assuming the 'can't attach' is referring to the zone.[06:48:11] <patrickmslattery> gist updated with json[06:48:32] <patrickmslattery> I get this unexpected token error[06:48:52] <patrickmslattery> If I take out the second NIC it works[06:49:30] <patrickmslattery> I have tried various changes for that second NIC, no go on any of them[06:51:58] <rmustacc> Looks like you have invalid json in there. You'll spot if it you cat jira.json into the json command.[06:52:16] <rmustacc> Specifically, you're missing a comma after the gateway line.[06:52:59] <patrickmslattery> something like: cat ~/mysql.json|json ?[06:53:58] <rmustacc> Yup[06:54:00] <patrickmslattery> Pretty sure I fixed that in the actual JSON file on the global zone, but let me try again...[06:54:37] <rmustacc> If you're getting an invalid json line, that's certainly the most useful way to figure out what's wrong.[06:55:53] <patrickmslattery> I'm getting all kinds of crap getting put into the JSON file, I updated the gist[06:56:17] <patrickmslattery> I've no idea where these extra chars are coming from[06:56:56] <patrickmslattery> Could it be my terminal type maybe? (Using xterm through SecureCRT on Windows)[06:57:32] <rmustacc> I'm not sure. I guess it's possible that some rather weird key codes are being sent through.[06:59:08] <patrickmslattery> It was the damn text editor! (NotePad++)[07:00:13] <patrickmslattery> I turned on show all chars and I notices that the spaces were somehow different on the bad lines, not tabs, but a different kind of space char, replaced them all with real spaces and now it works![07:00:46] <rmustacc> Ugh, that's annoying, but I'm glad it's working now.[07:01:13] <patrickmslattery> Now for the real test, doing a vmadm create...[07:05:05] <patrickmslattery> It's taking a very long time to provision compared to normal, shows in the vmadm list as "provisioning", is there some way I can see where it's at in the provisiong process?[07:06:49] <patrickmslattery> It timed out[07:06:50] <patrickmslattery> timed out waiting for /var/svc/provisioning to move for 604f880c-29db-432f-8c1e-608d39d9576d[07:07:48] <patrickmslattery> Based on a quick Google search that could be a DNS error[07:09:19] <patrickmslattery> Trying again with the Google DNS instead of internal DNS[07:13:31] *** nikolam has joined #smartos[07:15:12] <patrickmslattery> Using the Google DNS didn't work, trying with a static IP for the AD nic instead[07:17:15] <patrickmslattery> That seems to have worked.[07:18:24] <patrickmslattery> Looks like it might be this issue: https://github.com/joyent/smartos-live/issues/158[07:19:49] <patrickmslattery> Well I can ping both gateways so it looks like I'm in business :-)[07:20:25] <patrickmslattery> Thanks for the help rmustacc, didn't know about that json validation command[07:21:02] <rmustacc> It's acutally useful for a lot more than validation.[07:21:19] <patrickmslattery> Such as?[07:21:41] <rmustacc> It can be used to process json results and grab out portions of it.[07:21:54] <rmustacc> eg. you could do something like cat foo.json | json nics[07:22:06] <patrickmslattery> Ah, a smart grep[07:22:37] <rmustacc> Yeah, it basically lets you programatically access json fields[07:23:17] <patrickmslattery> Playing with it now, very slick![07:33:15] *** patrickmslattery has left #smartos[07:37:58] *** mamash has joined #smartos[07:41:50] *** cjones__ has quit IRC[07:41:51] *** mamash has left #smartos[07:45:58] *** nefilim has quit IRC[07:52:13] *** yruss972 has joined #smartos[07:53:38] *** yruss972 has quit IRC[07:54:50] *** yruss972 has joined #smartos[08:01:57] *** chakatz has joined #smartos[08:06:33] *** axonpoet has quit IRC[08:38:39] *** johnnuttall has joined #smartos[08:38:54] *** aortmann has joined #smartos[08:41:46] *** yruss972 has quit IRC[08:43:01] *** yruss972 has joined #smartos[08:47:51] *** FatDarrel has joined #smartos[09:07:19] *** groundwater has joined #smartos[09:10:36] *** tonist has joined #smartos[09:20:36] *** texarcana has quit IRC[09:22:37] *** texarcana has joined #smartos[09:25:46] *** alcir has joined #smartos[09:35:38] *** groundwater has quit IRC[09:36:14] *** yruss972_ has joined #smartos[09:37:27] *** yruss972 has quit IRC[09:41:31] *** yruss972_ has quit IRC[09:42:50] *** ktkNA is now known as ktk[09:47:17] *** nikolam has quit IRC[10:01:47] *** nikolam has joined #smartos[10:06:43] *** mgls has joined #smartos[10:13:15] *** mgls has quit IRC[10:15:01] *** scarcry has quit IRC[10:15:31] *** scarcry has joined #smartos[10:16:53] *** KermitTheFragger has joined #smartos[10:21:00] *** johnnuttall has quit IRC[10:21:32] *** andy_js has joined #smartos[10:37:08] *** johnnuttall has joined #smartos[10:44:26] *** FatDarrel has quit IRC[10:55:18] *** johnnuttall has quit IRC[10:57:57] *** mamash has joined #smartos[11:08:04] *** tonist has left #smartos[11:08:20] *** tonist has joined #smartos[11:12:10] *** FatDarrel has joined #smartos[11:21:53] *** johnnuttall has joined #smartos[11:27:10] *** mattconn_ has joined #smartos[11:28:23] *** mattcon__ has joined #smartos[11:28:30] *** mattconn_ has quit IRC[11:30:43] *** mattconnolly has quit IRC[11:34:49] *** johnnuttall has left #smartos[11:38:12] *** ivan\ has quit IRC[12:04:38] *** ivan\ has joined #smartos[12:05:20] *** nikolam has quit IRC[12:11:16] *** bluezenix has joined #smartos[12:20:16] *** chakatz has quit IRC[12:34:21] <alcir> why zones/dump has the checksum flag set to noparity[12:40:10] *** tonist has quit IRC[12:50:50] *** mattcon__ has quit IRC[12:51:26] *** mattconnolly has joined #smartos[12:55:26] *** mattconnolly has quit IRC[13:17:03] *** chakatz has joined #smartos[13:32:59] *** andy_js has quit IRC[13:33:27] *** szaydel has joined #smartos[13:42:27] *** andy_js has joined #smartos[14:09:01] <nshalman> Presumably so that the kernel doesn't have to calculate checksums when dumping at a panic.[14:15:32] *** tonist has joined #smartos[14:28:15] *** backjlack has joined #smartos[14:28:35] *** potatosalad has joined #smartos[14:39:36] *** bens1 has joined #smartos[14:40:18] *** elijah-mbp has quit IRC[14:43:34] <alcir> there is a place from which I can download previous release of smartos?[14:53:01] <nahamu> which one do you want?[14:53:27] <nahamu> as far as I know they haven't taken down any of the old images. The links are in the announcement emails.[14:55:03] *** chakatz has quit IRC[14:56:31] <nahamu> alcir: http://paste.ec/?0e0888ba8b720c85#72t9En3tZK3GLhZkBfRQPYg3VXZdI9tAWmb6bd3Oajo=[14:58:27] <alcir> very good[15:28:22] *** chakatz has joined #smartos[15:38:31] <ktk> question, I have to adjust a variable in a service configuration. do I have to export and re-import that or is there a direct way in doing that with svccfg?[15:39:58] <mamash> svccfg -s service_name 'setprop property_name = value'[15:40:25] <ktk> ok do I have to specify the propertygroup too?[15:40:40] <mamash> yeah[15:40:54] <mamash> sorry, should be property_group/property_name[15:41:11] <ktk> great tnx[15:41:19] <ktk> and I assume this is persistent like this?[15:41:25] <mamash> yes[15:41:31] <ktk> excellent[15:42:03] <mamash> might have to double quote the value if there are spaces or special chars[15:42:19] <ktk> ok[15:45:03] <ktk> mamash: worked fine, thanks![15:51:38] <nahamu> persistent as long as you're inside a NGZ.[15:51:46] <ktk> that's fine[15:51:53] <ktk> interesting thing is that the process fails now according to the log[15:52:04] <ktk> but when I start it from the shell the way it logs it everything works[15:52:17] <nahamu> environment variable issue?[15:52:34] <ktk> it seems to read it properly as it did adjust what I wanted[15:52:42] <ktk> (listening port)[15:54:06] <ktk> maintenance state means something didnt' work as expected so it stops forking?[15:56:00] <mamash> well, check the SMF log[15:56:05] <mamash> anything there?[15:56:45] <ktk> [ Sep 2 13:54:44 Executing start method ("/opt/local/sbin/varnishd -a 0.0.0.0:80 -l 64M -f /opt/local/etc/default.vcl"). ][15:56:48] <ktk> [ Sep 2 13:54:44 Method "start" exited with status 0. ][15:56:50] <ktk> [ Sep 2 13:54:44 Stopping because all processes in service exited. ][15:56:53] <ktk> [ Sep 2 13:54:44 Executing stop method (:kill). ][15:56:55] <ktk> [ Sep 2 13:54:44 Restarting too quickly, changing state to maintenance. ][15:57:04] <ktk> but if i copy that string varnish just starts well and listens on port 80[15:57:23] <ktk> does a service need to have special flags for listening on low ports?[16:00:31] <mamash> yes, if it runs as a non-privileged user[16:00:47] <ktk> ah that's probably the reason then[16:00:53] <ktk> it was 8080 so that will work[16:01:38] <mamash> need to add the net_privaddr privilege[16:01:46] <mamash> see e.g. here: http://www.vtoc.de/article/non-root_SMF_service[16:02:00] <mamash> (couldn't locate canonical docs on that this quick)[16:02:12] <ktk> great tnx[16:03:26] <ktk> can I add this with the set method from before or do I have to export/import the manifest?[16:03:36] <ktk> there is no priviledges block in my current one[16:04:05] <mamash> something like 'setprop start/privileges = astring: basic,net_privaddr'[16:04:14] <ktk> letme try that[16:12:41] *** yruss972 has joined #smartos[16:14:32] <yruss972> Hi guys- just noticed the php-mysql package has multiplied in 2013Q2. Was this on purpose?[16:14:59] <jperkin> yes[16:15:17] <jperkin> else you are restricted to a single version of mysql[16:15:58] <yruss972> you mean if you install both on the same machine? client and server?[16:17:08] <jperkin> no, php-mysql and p5-DBD-mysql etc can only build against one version of mysql at a time[16:17:18] <mamash> no, but if you prefer to use e.g. mysql 5.6 or 5.1 rather than the default 5.5[16:17:19] <jperkin> so we provide builds of them for each release of mysql/percona[16:17:41] <mamash> (or percona)[16:18:01] <yruss972> ok but a 5.5 client can talk to a 5.0 server - no?[16:18:40] <jperkin> possibly, you'd need to check ;)[16:18:42] <yruss972> anyway- I'm not harping just trying to fix chef which doesn't find php54-mysql anymore[16:19:07] <jperkin> just change it to the version you use, e.g. php54-mysql55[16:21:39] <yruss972> it's a lot of changes :? - could there be a default to the latest version at php54-mysql? I wonder what other pkg repos do[16:22:36] <jperkin> there's no easy way to do a default, and it would cause problems (what happens when some people want the default to change and others don't?)[16:22:46] <jperkin> better to be explicit about exactly what version you want[16:23:48] <jperkin> especially when there are competing forks and people prefer oracle/percona/mariadb/etc[16:28:40] <yruss972> Well- regarding the default, I don't think its a crime to say the default should be the latest stable version[16:28:40] *** backjlack has quit IRC[16:28:55] *** backjlack has joined #smartos[16:29:05] <yruss972> also the package names are somewhat redundant as it is[16:29:24] <yruss972> php54-mysql55-5.4.x[16:29:32] <yruss972> why 54 twice?[16:29:59] <jperkin> that's just an artefact of the version number being based on the php version number[16:30:16] <yruss972> right but take php54-http-1.7.5[16:30:19] <jperkin> for most other packages using the multi-version support they would be different[16:30:30] <yruss972> you could do php54-mysql-5.0[16:30:33] <yruss972> you could do php54-mysql-5.5[16:30:35] <yruss972> you could do php54-mysql-5.6[16:30:54] <yruss972> that would work much better with chef since you have to specify a version anyway[16:31:12] <yruss972> now I tell it to install php54-mysql v 5.4[16:31:28] <yruss972> and it takes the latest 5.4 package[16:32:07] <yruss972> with a naming scheme like php54-mysql-5.x I could say take v 5 and it would take 5.6[16:32:14] <yruss972> or I could specify 5.5[16:32:16] <yruss972> etc[16:32:42] <yruss972> not sure how that would work on your end[16:34:10] *** hile_ has joined #smartos[16:35:29] <jperkin> it's a simple change to amend the prefix, but it would then screw over any users who are already using the existing scheme[16:36:10] <hile_> I've got an interesting chicken/egg problem with SmartOS. If one PXE boots everything (say, there's existing pxe infrastructure in place), how does one seed the stuff in /opt/custom?[16:36:18] <yruss972> well I'm screwed already :/[16:39:15] <yruss972> If you can go in this direction in the future- it will work better for anyone using chef, otherwise we can manually change all our code to use new package names[16:39:20] <yruss972> or[16:39:46] <hile_> Mind you, I'm probably already doing things more differently than normal since I'm learning both chef and getting my head around smartos simultaneously.[16:40:38] *** tonist has quit IRC[16:40:46] <yruss972> well... no or really[16:43:24] <jperkin> I don't really like it, the 5.x isn't a version number, and it's going to be confusing at some point if tools are expecting it to be (what happens when there are multiple packages of that same 'version'?)[16:43:40] *** wiedi has quit IRC[16:44:51] *** yo61 has quit IRC[16:45:00] *** yo61 has joined #smartos[16:47:14] <Alasdairrr> Newer clients can talk to older servers[16:47:32] <Alasdairrr> If you're shipping multiple versions of mysql, php should be linked against the newest one[16:48:27] <yruss972> Alasdairrr: it could be an issue if you have mysql-server 5.0 installed on the same server as the php-mysql but otherwise I agree with you[16:50:13] <yruss972> jperkin: you could add another minor version number -5.0.1[16:50:38] <yruss972> or use these nb1 suffixes (which I never understood)?[16:51:23] *** nefilim has joined #smartos[16:51:56] *** mamash has left #smartos[16:52:44] <jperkin> no, it would already have a suffixed version, that of php's current version. 'php54-mysql-5.0' as the entire version string wouldn't work, I assumed you were just omitting them for brevity[16:53:21] <jperkin> nb? is a revision update unrelated to the upstream version, i.e. a patch or change in the build somewhere.[17:01:48] *** aortmann has quit IRC[17:02:11] *** leecallen has joined #smartos[17:05:00] *** leecallen35 has quit IRC[17:07:14] *** izaki has joined #smartos[17:08:59] *** alcir has quit IRC[17:18:20] *** dap has joined #smartos[17:24:08] *** cjones_ has joined #smartos[17:27:28] *** groundwater has joined #smartos[17:27:34] *** cjones_ has quit IRC[17:32:00] *** potatosalad has quit IRC[17:33:49] *** cjones_ has joined #smartos[17:37:38] *** yo61 has quit IRC[17:37:51] *** yo61 has joined #smartos[17:50:13] *** cjones_ has quit IRC[17:50:18] *** cjones__ has joined #smartos[17:55:15] *** nikolam has joined #smartos[17:57:31] *** yruss972 has quit IRC[17:57:39] *** bluezenix has quit IRC[17:59:15] *** benbangert has joined #smartos[18:01:33] *** dap has quit IRC[18:02:41] *** yruss972 has joined #smartos[18:06:28] *** potatosalad has joined #smartos[18:11:32] *** tonist has joined #smartos[18:14:21] *** bens1 has quit IRC[18:14:36] *** bens1 has joined #smartos[18:38:39] *** bluezenix has joined #smartos[18:39:50] *** yruss972 has quit IRC[18:43:08] <drscream> jperkin: thanks a lot for fixing the exim build :)[19:00:10] *** mamash has joined #smartos[19:03:24] *** chakatz has quit IRC[19:12:30] *** cjones__ has quit IRC[19:12:34] *** cjones_ has joined #smartos[19:19:18] *** KermitTheFragger has quit IRC[19:26:15] *** bluezenix is now known as vizanto[19:28:11] *** vizanto has quit IRC[19:40:36] *** tonist has quit IRC[19:42:18] *** mamash has left #smartos[19:44:31] *** mamash has joined #smartos[19:46:55] *** yruss972 has joined #smartos[19:51:14] *** yruss972 has quit IRC[19:52:25] <nahamu> is there a command line tool for blinking LEDs on bays connected to LSI 2308 cards?[19:52:59] <nahamu> I know Joyent added some stuff so that FMA can accurately report on drive bays in their configurations.[19:59:55] *** mamash has left #smartos[20:00:25] *** jgrafton has joined #smartos[20:00:31] *** Meths_ has joined #smartos[20:00:54] *** Meths has quit IRC[20:09:25] *** bleubeard has joined #smartos[20:21:35] <bleubeard> hello[20:29:11] *** Meths_ is now known as Meths[20:35:04] *** bens1 has quit IRC[20:44:22] *** tonist has joined #smartos[20:48:53] *** Dimeetrees has joined #smartos[21:00:32] <wesolows> nahamu: Maybe sort of -- it depends.[21:01:03] <nahamu> does that mean I can blink random lights?[21:01:08] *** mamash has joined #smartos[21:01:11] <wesolows> nahamu: if the backplane has an expander, then the LEDs are almost certainly controlled by the SES target in the expander, in which case they already will work.[21:01:18] <wesolows> no, that's not what it's for.[21:02:02] <wesolows> There is an FMA module that will illuminate the LEDs, using either SES or (in the case of Joyent-branded chassis with direct-attach LSI HBAs) SGPIO via an mpt_sas ioctl, when a disk fault is detected.[21:02:38] <wesolows> If you want to use that same mechanism manually, you may be able to do so using /usr/lib/fm/fmd/fmtopo -- look at the "method" and "indicator" stuff, and use the source because this is not a Public interface.[21:03:22] <nahamu> I'll poke around. thanks.[21:03:22] <wesolows> It may be possible to dork with a third LED (perhaps "ready to remove" or "identify", depending on your chassis) via that same mechanism, iff the LEDs are under SES target control and the target presents them to the system.[21:03:40] <wesolows> The Activity LED is almost never software-visible at all on any system.[21:04:12] <wesolows> The first thing to figure out in any case is whether you are SES or DA.[21:04:26] <nahamu> I suspect it's DA.[21:04:35] <nahamu> can prtconf help me figure that out?[21:04:40] <wesolows> You can write simple C programs using either libses or the ioctls in mpt_sas to control what LEDs it is possible to control...[21:04:56] <wesolows> maybe? You'd probably have an "enclosure services" SCSI target if it's under SES control.[21:05:22] <wesolows> fmtopo -V can tell you also, if you know what to look for.[21:05:41] <LeftWing> Man, LEDs.[21:06:11] <wesolows> I strongly discourage fucking with the Fault/Fail LED though; that's FMA's job. The locate/identify one, though, would be nice to be able to toggle manually.[21:06:24] <LeftWing> You can do that now![21:06:30] <wesolows> oh, you got that working?[21:06:32] <wesolows> cool.[21:06:42] <LeftWing> Assuming you have a bay=N?indicator=ident[21:06:44] <nahamu> LeftWing: in C or from the command line?[21:06:47] <wesolows> yeah.[21:07:00] <wesolows> This is the magic fmtopo invocation I can never remember.[21:07:16] <LeftWing> You can use fmtopo with "fmtopo -P facility.mode=uint32:0x1 <fmri>" to turn it on[21:07:44] * wesolows thinks there's a good reason this interface isn't Public :)[21:07:49] <nahamu> indeed.[21:08:07] *** tonist has quit IRC[21:08:12] <wesolows> We should really beef up diskinfo some more.[21:08:32] <nahamu> what would I see in fmtopo -V if I had SES?[21:08:34] <LeftWing> Yeah, I think I assigned that to myself[21:08:57] <wesolows> nahamu: You would see some properties related to SES in each bay/disk :)[21:09:15] <LeftWing> Stuff like... hc://:product-id=LSI-CORP-SAS2X28:server-id=:chassis-id=5003048001eeb5ff:serial=YGH80WPD:part=HITACHI-HUS723020ALS640:revision=A222/ses-enclosure=0/bay=0/disk=0[21:09:39] <nahamu> nope, this box is clearly DA.[21:10:06] <LeftWing> can you gist ... "mdb -k -e '::mptsas -t'" ?[21:11:10] <nahamu> http://paste.ec/?a9b1d470bb484a39#fH8HmGtjBj1vi2yaf15x1dnDaoyorEc8eIo+xynGTTs=[21:11:43] <LeftWing> heh... weird. SATA disks?[21:11:43] <jzu_> hmm, is Cloud Analytics available for generic SmartOS hosts?[21:11:50] <nahamu> yes, sadly.[21:11:52] <jzu_> or is CA something that's only part of SDC?[21:12:11] <LeftWing> CA is a part of SDC, jzu_.[21:12:31] <jzu_> LeftWing: but it's not available seperately?[21:12:37] <LeftWing> It's an upstack service for aggregating the DTrace/kstat/etc output from _many_ SmartOS hosts. Certainly the underlying tools are in the OS thoughl[21:13:09] <jzu_> yeah, would have liked to try doing heatmaps of only one host running SmartOS[21:13:23] <LeftWing> You could use terminal-heatmap :P[21:13:31] <jzu_> :p[21:13:38] *** Dimeetrees has quit IRC[21:13:48] <LeftWing> https://github.com/jclulow/terminal-heatmap :P[21:14:19] <LeftWing> Brendan also has some post-processing tools to make heatmaps from DTrace output, etc... https://github.com/brendangregg/HeatMap[21:14:34] *** tonist has joined #smartos[21:15:29] <ktk> damn how could I live without SMF[21:15:40] <nahamu> ktk: much less comfortably.[21:16:14] <ktk> true[21:16:21] <ktk> and why does everyone say it's complicated[21:16:24] <ktk> it's not[21:16:27] <nahamu> XML.[21:16:35] <LeftWing> The implementation underneath is a touch complicated.[21:16:40] <ktk> yeah was about to say one might argue about xml[21:17:35] <LeftWing> The data model (including snapshots, and inheritance, etc) is also kind of complicated. But you mostly don't need to worry about that stuff to do basic services.[21:17:47] <ktk> still have to read about some of the config options though. is there something like the ZFS manual from former sun for SMF?[21:17:56] <ktk> the ZFS manual is like the best documentation I've ever seen[21:18:18] <ktk> LeftWing: ah didn't even see that yet[21:19:02] <LeftWing> If you're doing single-instance services and not really using configuration in the service itself, other than selecting what script to use, etc, then you won't really be exposed to that kind of complexity.[21:19:57] <nahamu> LeftWing: before I try to turn on one of the LEDs like that, how would I turn it back off? 0x0?[21:20:07] <ktk> ok FreeBSD is still much better than all linux stuff I worked with but this zone stuff is the most elegant implementation I've seen so far[21:20:07] <LeftWing> nahamu: Indeed.[21:20:27] <LeftWing> nahamu: Surely you don't have bay/indicator nodes, though?[21:20:36] *** mamash has left #smartos[21:21:13] <nahamu> I do not. So I can't do that, correct?[21:21:25] *** julianduque has joined #smartos[21:21:26] *** julianduque has joined #smartos[21:21:31] <ktk> only thing I'm a bit surprised about is the lack of IPv6 stuff, you guys in the states obviously still have too many v4 addresses :)[21:21:51] <LeftWing> nahamu: Correct. The facility.mode property you'd be setting would be on a facility node. i.e. "bay=0?indicator=fail", etc[21:21:58] <nahamu> ah[21:22:33] <LeftWing> ktk: The OS is fully IPv6 capable. The business driver for us to finish IPv6 configuration support in the base platform has not yet been there. It'll happen sooner or later.[21:22:44] <nahamu> LeftWing: so how do you do lights on DA drives[21:22:55] *** tonist has quit IRC[21:23:11] <LeftWing> nahamu: So, that dcmd I had you dump out of mptsas shows you "enclosure" and "slot" numbers as mpt_sas perceives them[21:23:25] <nahamu> yup, I noticed that.[21:23:28] <ktk> LeftWing: looking forward to that, tried playing with it manually but it's a bit PITA[21:23:39] <LeftWing> ktk: IPv6 is always a PITA :)[21:25:15] <LeftWing> nahamu: So, assuming your mpt_sas isn't broken, and your manufacturing process plugs the same DA disk into the same HBA port every time[21:25:46] <LeftWing> You can create an XML mapping with those encl/slot numbers that tells FMA about the disk bays in the server[21:25:49] <LeftWing> See: https://github.com/joyent/illumos-joyent/blob/master/usr/src/lib/fm/topo/maps/Joyent%2CJoyent-Compute-Platform-1101/Joyent-Compute-Platform-1101-disk-hc-topology.xmlgenksh[21:26:39] <LeftWing> You also need the path to the "devctl" node for the particular mpt_sas instance.[21:26:49] <LeftWing> (so that FMA can send ioctls)[21:27:44] <LeftWing> The way we _select_ this static map is based on the SMBIOS "Product" field...[21:27:46] <LeftWing> ... and this file: https://github.com/joyent/illumos-joyent/blob/master/usr/src/lib/fm/topo/maps/i86pc/i86pc-legacy-hc-topology.xml#L155-L165[21:28:16] *** bens1 has joined #smartos[21:28:33] <LeftWing> So in the boxes we have manufactured at Joyent, we write a product identifier into the SMBIOS Product field like "Joyent-Compute-Platform-1101". This explicitly calls out a particular physical layout of parts (including disk bays, etc)[21:30:45] <nahamu> LeftWing: so is that XML file specifically for teaching FMA how to find the DA drives because everything "just works" with the drives that are in SES enclosures?[21:30:59] <nahamu> (FSVO "just works")[21:31:09] <LeftWing> Correct. SES enclosures are generally self-identifying[21:31:13] <nahamu> (where there's firmware, nothing "just works", etc.)[21:31:44] <LeftWing> Not just firmware, here -- also shoddy design and manufacturing, leading to stupid "label" strings.[21:32:05] <nahamu> jost "Joyent-Compute-Platform-1101" match any of the output in sysinfo?[21:32:06] <LeftWing> The physical box may have "BAY 1" printed on the same thing that is identified as "Front Bay 0" via SES.[21:32:08] <nahamu> *does[21:32:27] <LeftWing> It'll be in "Product" in there. It comes out of "smbios -t 1"[21:32:30] <nahamu> yeah, I wanted to audit what the OS thinks is where compared to the labels.[21:32:46] <LeftWing> If you run "smbios -t 1" now, you'll see what we believe your system to be. :)[21:33:03] <nahamu> which in my case shows the motherboard model[21:33:37] <LeftWing> Right. That's what we had, too.[21:33:53] <LeftWing> I made a DOS boot image with an AMI tool on it to write new SMBIOS values.[21:34:02] <LeftWing> And then we pushed that back into the manufacturing process.[21:34:50] *** cjones_ has quit IRC[21:34:54] *** cjones__ has joined #smartos[21:35:39] <LeftWing> I hope to write a blog post about the end-to-end process of getting the disks to light up some time.[21:36:03] <nahamu> so in theory I could generate my own xml file to enumerate 8 drive bays, modify the i86pc-legacy-hc-topology.xml file to pull it in, restart the appropriate service, and then perhaps toggle the lights with fmtopo[21:36:52] <LeftWing> Well, so the second sticking point here is that the "sas address" field for your disks seems mostly to be "0". Which is not ideal.[21:36:59] <nahamu> (and of course it's a private interface, so be careful)[21:38:44] <nahamu> yeah, I found that odd given that the other drive seemed to generate an appropriately complicated address.[21:39:57] <LeftWing> I can also just give you a small C program that ioctls directly into the HBA, if you're just keen to turn a light on.[21:40:38] <nahamu> also, given the enclosure and slot numbers and which one has the outlier, there's clearly a disconnect between the sticker numbers and which port on the card is connected.[21:40:56] <LeftWing> Well there's no guarantee that the encl/slot numbers _mean_ anything[21:40:58] *** tonist has joined #smartos[21:41:03] <LeftWing> They're just hopefully a unique mapping to each bay :)[21:41:16] <nahamu> they shouldn't change across a reboot, should they?[21:41:45] <LeftWing> Well, I can't make any promises, but they should not. Certainly if they do then your mpt_sas HBA/firmware is hilariously off-colour.[21:43:07] <nahamu> f'ing firmware... ;)[21:43:24] <nahamu> should be a sane version, and it's a 2308 chip.[21:43:46] <nahamu> SAS9207-8i card[21:44:31] <nahamu> LeftWing: if you already have the c program, I'm sure it would be both useful and instructive.[21:44:52] <LeftWing> I've just tried looking for it, and I can't find it right now. But, it's basically this: https://github.com/joyent/illumos-joyent/blob/master/usr/src/lib/fm/topo/modules/common/fac_prov_mptsas/fac_prov_mptsas.c#L75-L128[21:45:26] <LeftWing> You should be able to lift that out, cleave out the libtopo-specific bits and just do the ioctl.[21:45:53] <LeftWing> You'll be wanting a LED type of ... MPTSAS_LEDCTL_LED_IDENT, I think.[21:46:34] <LeftWing> I have to head out, now. Let me know if you have questions and I'll try and answer them when I'm around.[21:46:52] <nahamu> thanks![21:48:06] *** bens1 has quit IRC[21:48:59] *** tonist has quit IRC[21:50:08] *** mamash has joined #smartos[21:54:49] *** potatosalad has quit IRC[21:58:31] *** mamash has left #smartos[22:06:45] *** tonist has joined #smartos[22:14:31] *** tonist has quit IRC[22:41:50] *** cjones__ has quit IRC[22:42:28] *** ktk is now known as ktkNA[22:43:36] *** bens1 has joined #smartos[22:48:00] *** tonist has joined #smartos[22:48:15] *** bens1 has quit IRC[22:49:37] *** tbatchelli_m has joined #smartos[22:51:58] *** potatosalad has joined #smartos[22:55:11] *** mamash has joined #smartos[23:00:27] *** nathan7 has quit IRC[23:05:23] *** mamash has left #smartos[23:13:43] *** nathan7 has joined #smartos[23:15:36] *** tonist has quit IRC[23:16:54] *** tbatchelli_m has quit IRC[23:18:15] *** tbatchelli_m has joined #smartos[23:19:02] *** zr0 has quit IRC[23:25:58] *** andy_js has quit IRC[23:28:14] *** mattconnolly has joined #smartos[23:35:16] *** groundwater has quit IRC[23:40:31] *** mattconnolly has quit IRC[23:41:06] *** mattconnolly has joined #smartos[23:43:54] *** mattconnolly has quit IRC[23:44:09] *** mattconnolly has joined #smartos[23:44:43] *** mattconnolly has quit IRC[23:45:16] *** mattconnolly has joined #smartos[23:49:19] *** mattconnolly has quit IRC[23:49:54] *** zr0 has joined #smartos[23:56:45] <jelmd> LeftWing: ping[23:58:38] *** cjones_ has joined #smartos