September 17, 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 17, 2013 [00:07:56] *** jaimef has quit IRC[00:12:31] *** volci has joined #smartos[00:13:02] *** jaimef has joined #smartos[00:26:53] <yo61> So, I want to copy a wrapper script and a config file from files/ to the package directory[00:27:28] <yo61> I've put ${CP} commands in a pre-configure: section[00:27:54] <yo61> However, I'm using NO_CONFIGURE= yes so it seems the pre-configure hook is not firing[00:33:09] <yo61> I can put the ${CP} commands in a pre-install: section, but the substitution doesn't seem to work if I do that (not that I'm sure it's working in the pre-configure: hook)[00:33:32] *** andy_js has quit IRC[00:37:13] *** mary5030 has quit IRC[00:37:28] *** jsm has quit IRC[00:37:33] *** miine has quit IRC[00:45:49] *** arekinath_ is now known as arekinath[00:45:53] *** arekinath has quit IRC[00:45:54] *** arekinath has joined #smartos[00:46:23] *** jim80net has quit IRC[00:54:02] <yo61> OK, figured it out.[00:54:16] <yo61> However, how can I perform substitution on the SMF manifest?[01:11:10] *** ilovezfs has quit IRC[01:12:58] *** backjlack has quit IRC[01:14:08] *** tbatchelli_m has quit IRC[01:17:11] *** ilovezfs has joined #smartos[01:20:50] <yo61> Yay, finished something and it works: https://github.com/robinbowes/pkgsrc-elasticsearch[01:23:45] <rmustacc> Nice.[01:30:18] <bixu> yo61: Yeah - nice work. After doing this scores of times, I've settled (personally) on avoiding wrapper scripts in favor of just passing a little command-line argument in the SMF manifest. Often will have the config file referenced there.[01:30:42] <bixu> Not saying that's the Right Way™, just what I've settled on for the most part.[01:31:52] <bixu> Part of that may be because we use Chef and Eric Saxby's SMF cookbook/provider: https://github.com/livinginthepast/smf[01:32:56] <bixu> (Ironically, the README for Eric's cookbook is probably one of the most concise overviews of SMF concepts - he and I hashed it out over the last year or so at two different companies.)[01:32:59] <yo61> bixu: a command-line argument in the manifest couldn't be optional, could it?[01:33:17] <bixu> yo61: I'm not sure what you mean.[01:33:59] <yo61> Well, all the wrapper does is [ -f /path/to/local.conf ] && . /path/to/local.conf[01:34:24] <yo61> How would you do that in a manifest?[01:34:29] <bixu> You can do that as your exec method in the manifest, I think.[01:34:39] <bixu> I believe it just shells out.[01:35:02] *** volci has quit IRC[01:35:08] <bixu> If you don't use the test, the method will just go to maintenance because the file doesn't exist.[01:35:16] <yo61> Right[01:35:18] <bixu> You can also specify that conf file as a dependency for the service.[01:36:09] <bixu> Do you want the service to configure itself differently depending on whether the that particular conf file exists?[01:36:32] <yo61> That config file sets various env vars to configure the jvm[01:37:18] <bixu> What I'm getting at is that you don't need to test for the existence of the file because you can have SMF do that for you by making the file a dependency.[01:37:50] <yo61> But a dependency must exist, mustn't it?[01:37:54] <yo61> That config is optional[01:38:02] <bixu> I see.[01:39:47] <yo61> The idea being that the daemon will run out-of-the-box[01:39:58] <bixu> I still think (IIRC) that you can put specify that whole command line as the service method.[01:40:12] <yo61> I may give it a go tomorrow[01:40:20] <yo61> I'm still learning this stuff[01:40:33] <bixu> Yeah - SMF took me a good long while to grok.[01:40:44] <bixu> And I'm still learning about little corners of it.[01:43:03] <yo61> I like the general approach - I'm a big fan of the daemontools/runit approach[01:44:43] <bixu> Agreed. I don't know the history of these kinds of service management tools, but I have to say that SMF feels the most..."industrial grade"...to me. Once you learn it, you won't want anything else.[01:45:18] <yo61> Well, smf is possibly over-engineered, IMHO[01:45:26] <bixu> And if you are using Chef, Eric's cookbook takes a lot of the pain out of xml wrangling. I think there are also some node-based tools for turning JSON into an SMF manifest.[01:45:42] <bixu> That was also my initial reaction.[01:45:49] <bixu> YMMV of course.[01:45:57] <yo61> manifold is helpful[01:45:58] <tjfontaine> bixu: btw, things are going swimingly now?[01:46:09] <bixu> tjfontaine: WRT?[01:46:25] <tjfontaine> your chef and manta stuff, I just realized EWRONGCHAN[01:46:32] <bixu> Haha.[01:46:38] <bixu> Yes - sorted.[01:46:48] <yo61> The bit I like most is not having to mess with pid files etc.[01:46:49] <bixu> Still think your ideas about error handling are apropos.[01:46:54] <tjfontaine> nod[01:46:57] <bixu> yo61: AMEN[01:47:27] <bixu> Management via pidfile when we have better tools makes me depressed.[01:47:35] <yo61> No daemon mode, just launch in the foreground and let smf/runit/daemontools manage it[01:48:08] <bixu> It's like living in the 21st century.[01:48:37] <yo61> Anyway, time for me to hit the hay[01:48:59] <nahamu> or at least if it automatically backgrounds itself, SMF can track any processes it forks and kill *all* of them at shutdown time[01:50:06] <bixu> nahamu: Yeah. Funnily enough, although a lot of rubyists don't know much about SmartOS/SMF, we've found (in a rails shop) that SMF is pretty ideal for managing stuff like Unicorn or Sidekiq.[01:50:51] <nahamu> I'm not surprised.[01:51:43] <bixu> Partly because a lot of Rails stuff get's rapidly prototyped, and so there's not always good daemonization/child management. SMF happens to be good at that.[01:54:57] <richlowe> I'd say demonizing rails is something we're pretty good at.[01:55:44] <bixu> richlowe: Very funny.[01:56:07] *** tbatchelli_m has joined #smartos[01:56:32] *** tbatchelli_m has quit IRC[01:56:50] <bixu> Rails get's a bad name sometimes, but having worked in rails shops for a few years now, I definitely see where it shines.[01:56:53] <bixu> :)[01:57:06] *** tbatchelli_m has joined #smartos[02:16:18] *** tbatchelli_m has quit IRC[02:38:46] *** khushildep has quit IRC[02:39:28] *** julianduque has quit IRC[02:40:43] *** julianduque has joined #smartos[03:02:00] *** mary5030 has joined #smartos[03:13:38] *** dap has quit IRC[03:13:45] *** Thrae has quit IRC[03:17:40] *** Thrae has joined #smartos[03:19:02] *** julianduque has quit IRC[03:28:47] *** mmalecki is now known as mmalecki[zzz][03:39:57] *** bixu has quit IRC[03:41:50] *** pringlescan1 has quit IRC[03:50:36] *** volci has joined #smartos[03:57:06] *** mary5030 has quit IRC[03:57:40] *** mary5030 has joined #smartos[03:59:25] *** potatosalad has quit IRC[03:59:44] *** mary5030 has quit IRC[04:06:16] *** potatosalad has joined #smartos[04:06:30] *** nefilim has joined #smartos[04:11:18] *** potatosalad has quit IRC[04:15:51] *** jellydonut has quit IRC[04:23:45] *** kevinykchan has joined #smartos[04:27:53] *** mary5030 has joined #smartos[04:31:07] *** saskaloon has quit IRC[04:32:38] *** saskaloon has joined #smartos[04:47:33] *** wolstena has quit IRC[04:47:55] *** Alek has quit IRC[04:54:01] *** wolfeidau has quit IRC[04:54:13] *** wolfeidau has joined #smartos[05:16:15] *** jim80net has joined #smartos[05:17:13] *** szaydel has quit IRC[05:23:05] *** pringlescan has joined #smartos[05:27:53] *** szaydel has joined #smartos[05:30:23] *** mary5030 has quit IRC[05:32:45] *** julianduque has joined #smartos[05:43:13] *** cjones_ has quit IRC[06:05:03] *** yruss972 has joined #smartos[06:06:12] *** yruss972 has quit IRC[06:11:56] *** pringlescan has quit IRC[06:15:48] *** yruss972 has joined #smartos[06:35:16] *** julianduque has quit IRC[06:37:35] *** nefilim has quit IRC[06:38:26] *** potatosalad has joined #smartos[06:39:56] *** szaydel has quit IRC[07:09:52] *** potatosalad has quit IRC[07:23:08] *** ivan\ has quit IRC[07:26:18] *** ivan\ has joined #smartos[07:38:18] *** vizant0 has joined #smartos[08:00:00] *** sjorge has quit IRC[08:02:58] *** khushildep has joined #smartos[08:04:48] *** jzu__ has joined #smartos[08:05:05] *** benbange1t has joined #smartos[08:05:09] *** prasm_ has joined #smartos[08:05:20] *** cmang_ has joined #smartos[08:05:39] *** kwork_ has joined #smartos[08:05:59] *** calmh- has joined #smartos[08:06:41] *** khushildep_ has joined #smartos[08:07:03] *** khushildep has quit IRC[08:07:03] *** khushildep_ is now known as khushildep[08:08:53] *** rodgort` has joined #smartos[08:09:09] *** wramthun has quit IRC[08:09:10] *** prasm has quit IRC[08:09:11] *** kwork has quit IRC[08:09:16] *** jelmd has quit IRC[08:09:17] *** benbangert has quit IRC[08:09:17] *** rodgort has quit IRC[08:09:19] *** hij1nx has quit IRC[08:09:20] *** scarcry has quit IRC[08:09:23] *** calmh has quit IRC[08:09:25] *** jzu_ has quit IRC[08:09:25] *** cmang has quit IRC[08:09:26] *** ismell has quit IRC[08:09:27] *** nikolam has quit IRC[08:09:39] *** ismell_ has joined #smartos[08:09:39] *** ismell_ is now known as ismell[08:11:38] *** hadret has joined #smartos[08:12:57] *** scarcry has joined #smartos[08:13:14] *** hij1nx has joined #smartos[08:13:25] *** wramthun has joined #smartos[08:14:18] *** trentster_ has joined #smartos[08:14:33] *** wesolows_ has joined #smartos[08:15:02] *** griffin__ has joined #smartos[08:15:35] *** jones_d_ has joined #smartos[08:15:41] *** Xenith_ has joined #smartos[08:16:14] *** Prudhvi_ has joined #smartos[08:17:39] *** aszeszo` has joined #smartos[08:18:16] *** trentster has quit IRC[08:18:18] *** jamesd_ has quit IRC[08:18:21] *** Xenith has quit IRC[08:18:28] *** jones_d has quit IRC[08:18:30] *** wesolows has quit IRC[08:18:33] *** griffin_ has quit IRC[08:18:34] *** jihyun has quit IRC[08:18:34] *** Prudhvi has quit IRC[08:18:34] *** socketwiz has quit IRC[08:18:37] *** aszeszo has quit IRC[08:18:37] *** Raynos has quit IRC[08:18:38] *** jihyun_ has joined #smartos[08:18:39] *** trentster_ is now known as trentster[08:20:58] *** defel has joined #smartos[08:22:06] *** ssm_ has joined #smartos[08:22:37] *** prasm has joined #smartos[08:23:55] *** prasm_ has quit IRC[08:23:56] *** Licenser has quit IRC[08:23:56] *** defel_ has quit IRC[08:24:16] *** jaimef has quit IRC[08:24:17] *** ssm has quit IRC[08:24:35] *** scarcry has quit IRC[08:24:35] *** utlemming has quit IRC[08:24:41] *** e^ipi has quit IRC[08:24:43] *** badboy_ has quit IRC[08:24:44] *** scarcry has joined #smartos[08:25:05] *** jaimef_ has joined #smartos[08:25:06] *** e^ipi has joined #smartos[08:25:07] *** Licenser has joined #smartos[08:25:28] *** badboy has joined #smartos[08:26:15] *** badboy is now known as Guest88493[08:28:14] *** tonist has joined #smartos[08:33:05] *** sjorge has joined #smartos[08:50:53] *** tonist has quit IRC[08:53:12] *** porkotron has joined #smartos[08:53:12] *** utlemming has joined #smartos[08:53:12] *** jamesd has joined #smartos[08:53:50] *** jihyun_ has quit IRC[08:53:50] *** jihyun_ has joined #smartos[08:55:46] *** nikolam has joined #smartos[09:10:15] *** wiedi has joined #smartos[09:10:43] *** aszeszo` has quit IRC[09:11:13] *** aszeszo has joined #smartos[09:14:20] *** alcir has joined #smartos[09:20:01] *** texarcana has quit IRC[09:20:44] *** kwork_ is now known as kwork[09:21:49] *** nikolam has quit IRC[09:22:06] *** texarcana has joined #smartos[09:44:38] *** ktkNA is now known as ktk[09:51:59] *** Guest88493 is now known as badboy[09:52:29] *** badboy is now known as Guest39553[09:52:54] *** Guest39553 is now known as badboy_[09:59:36] *** mattconnolly has quit IRC[10:04:58] *** bens1 has joined #smartos[10:06:08] *** kevinykchan has quit IRC[10:09:14] *** aortmann has joined #smartos[10:10:16] <aortmann> hi guys, my smartos server crashed (power blackout). Now I got the error message short read at boot... :/ any one got an idea?[10:15:34] *** Raynos has joined #smartos[10:20:44] *** mmalecki[zzz] is now known as mmalecki[10:28:18] *** nikolam has joined #smartos[10:29:43] *** the_TORmentor has joined #smartos[10:35:05] *** andy_js has joined #smartos[10:40:56] *** the_TORmentor is now known as The_Tormentor[10:41:24] *** _joes_ has joined #smartos[11:02:58] *** ktk has quit IRC[11:07:44] *** backjlack has joined #smartos[11:14:24] *** W0rmDr1nk has joined #smartos[11:15:33] *** vizant0 has quit IRC[11:21:44] *** estibi has joined #smartos[11:38:40] *** mamash has joined #smartos[12:03:38] *** alcir has quit IRC[12:04:56] *** alcir has joined #smartos[12:15:27] *** vizant0 has joined #smartos[12:31:59] *** The_Tormentor has quit IRC[12:37:02] *** xenol has joined #smartos[12:42:05] *** thorfinn has joined #smartos[12:47:22] *** vizant0 has quit IRC[12:55:52] *** vizant0 has joined #smartos[13:33:24] *** szaydel has joined #smartos[13:33:41] *** andy_js has quit IRC[13:37:31] *** W0rmDr1nk has quit IRC[13:37:59] *** W0rmDr1nk has joined #smartos[13:46:25] *** andy_js has joined #smartos[13:46:43] <xenol> Hi, I know that if was surely asked gazillion times before me, but when is the plan to implement feature that vmadmd listens to the network and I can vmadm send vm directly to it?[13:46:51] <xenol> s/if/it[13:47:28] <jperkin> I don't know of a plan to implement that[13:48:55] <xenol> https://github.com/joyent/smartos-live/blob/master/src/vm/README.migration might be hinting something like that[13:49:06] <xenol> "vmadm send ${UUID} ${REMOTE_VMADMD_IP} ${PORT} && vmadm delete ${UUID}"[13:50:05] *** nikolam has quit IRC[13:51:50] <yo61> Why would you want/need that?[13:52:27] <xenol> migration of VMs between smartos nodes[13:52:50] <jperkin> you can of course do something very similar using ssh as the tunnelling method, doing it directly to a vmadmd instance would require a bunch of authentication issues[13:52:51] <xenol> at the moment, I am using either zfs send/recv + copying some files from /etc/zones or vmadm send[13:53:03] <yo61> I was meaning why would you need to do that directly without using ssh ?[13:53:16] <xenol> jperkin: I am using that method[13:54:30] *** jelmd has joined #smartos[13:55:01] <xenol> I like to lock my smartos nodes down (more security.)[13:55:16] <xenol> though, the vmadmd authentication is a good point :)[13:58:34] <yo61> Hmm, vmadm send/receive not in man pages[13:59:48] <jperkin> it's not yet supported[14:00:01] <jperkin> we don't want to give the impression it is anything other than experimental[14:00:09] <yo61> xenol: found this in my searching: https://github.com/bixu/vmsend[14:00:14] <yo61> jperkin: got it[14:02:33] <xenol> yo61: it is not in man page, because it is experimental and not documented intentionally.[14:03:00] *** nikolam has joined #smartos[14:03:02] *** jellydonut has joined #smartos[14:15:30] *** xenol has quit IRC[14:15:38] *** xenol has joined #smartos[14:18:52] *** Shadow_aok has joined #smartos[14:19:00] <Shadow_aok> hi guys[14:24:46] <Shadow_aok> I have an issue when trying to create a VM (SmartOS 20130822).[14:24:46] <Shadow_aok> vmadm create -f test.json[14:24:47] <Shadow_aok> timed out waiting for /var/svc/provisioning to move for 88705d28-ece5-4eb4-b58f-187d1316aba4[14:24:55] <yo61> Aha![14:24:56] <Shadow_aok> Here's the json file : http://pastebin.com/6b1NDEkP[14:25:11] <Shadow_aok> yo61> you have the same issue ?[14:25:39] <Shadow_aok> I tried with external as nic tag but I got an error message "invalid nic tag external"[14:26:10] <yo61> I had the same error, but it was for a native zone and was cause by some obscure locking bug when some zfs property was enabled on the zones zpool[14:26:23] <Shadow_aok> the json network parameters are valid and permit net access[14:26:34] *** jim80net has quit IRC[14:26:36] <Shadow_aok> how did you fix it ? Maybe it"s the same problem.[14:27:12] <yo61> one sec[14:27:18] *** hotbox_ is now known as hotbox[14:27:27] <yo61> (I doubt it)[14:27:57] <Shadow_aok> well, at least if i ran into the issue with a native zone, i'll know how to fix this one[14:28:26] <yo61> https://github.com/joyent/smartos-live/issues/261[14:28:50] <yo61> The fix was: zfs set nbmand=off zones[14:29:01] <yo61> But this is probably not related to your issue[14:29:05] <LnxTx> try remove: "default-gateway": "192.168.3.1",[14:29:20] *** jim80net has joined #smartos[14:29:31] <jperkin> the first thing I'd always check is any changes you have made to the default install, and then whether you can create other images.[14:30:00] <Shadow_aok> ok[14:30:25] <Shadow_aok> well, i can post my config too, but i haven't made any changes otherwise[14:30:35] <jperkin> there's no such option as default-gateway fwiw, perhaps you were looking for 'gateway' ?[14:30:39] <Shadow_aok> it's running on a 3 disks raid 5 smartos created[14:30:56] <Shadow_aok> default-gateway is in the documentation[14:31:11] <Shadow_aok> http://wiki.smartos.org/display/DOC/How+to+create+a+KVM+VM+%28+Hypervisor+virtualized+machine+%29+in+SmartOS[14:32:16] <jperkin> ok, someone needs to update the manual page then ;)[14:32:55] <jperkin> log.warn('DEPRECATED: default_gateway should no longer be used, '[14:32:56] <jperkin> + 'instead set one NIC primary and use nic.gateway.');[14:33:21] <Shadow_aok> thanks[14:33:23] <jperkin> I'd always go with what the manual page and code say, the wiki can be outdated and is never authoritative[14:36:41] <Shadow_aok> Ok, here i go : http://pastebin.com/dkiMQ9KZ[14:36:43] <Shadow_aok> same error[14:37:06] <Shadow_aok> damn, i missed the primary[14:41:26] <LnxTx> if you using disk image, you shouldn't use disks.0.size?[14:43:01] <Shadow_aok> well, it's disks.0.size in the json[14:43:12] <Shadow_aok> it's the first and only entry under disks[14:45:08] <Shadow_aok> Not working either : http://pastebin.com/T3UJbBuC[14:45:24] <LnxTx> show us: imgadm get d2ba0f30-bbe8-11e2-a9a2-6bc116856d85[14:45:48] <LnxTx> i think disk size is larger than 1024 MB[14:46:01] <LnxTx> disk image size[14:46:14] <Shadow_aok> http://pastebin.com/kxNVF3At[14:46:21] <jperkin> also check the logs under /var/log/vm[14:46:41] <Shadow_aok> damned, 1024 instead of 10240[14:46:44] <LnxTx> "image_size": 16384[14:48:10] <Shadow_aok> ok,trying again with 20480 as disk size[14:50:20] *** yruss972 has quit IRC[14:53:20] *** yruss972 has joined #smartos[14:53:32] <Shadow_aok> still the same error[14:55:44] <LnxTx> hmm, show logs: /zones/<UUID>/root/tmp/vm.log[14:59:50] *** khushildep_ has joined #smartos[15:00:38] *** khushildep has quit IRC[15:04:15] *** khushildep has joined #smartos[15:04:43] *** khushildep_ has quit IRC[15:06:26] <Shadow_aok> i need to create it again, i already removed it[15:07:10] *** mamash has left #smartos[15:07:39] <Shadow_aok> i updated the wiki to remove default-gateway in the json example[15:10:02] <Shadow_aok> vm.log : http://pastebin.com/0JkX2fTn[15:10:21] <Shadow_aok> (creation isn't finished yet)[15:10:30] <Shadow_aok> i'll update the pastebin if needed[15:10:32] <jperkin> I see the same failure, fwiw[15:11:34] <Shadow_aok> looks like kvm support is missing[15:11:51] <jperkin> oh, right, heh[15:12:07] *** W0rmDr1nk has quit IRC[15:12:13] <jperkin> that may affect provision speed and hit the timeout, qemu will still work but _very_ slowly[15:12:30] <Shadow_aok> sysinfo shows "VM Capable": true[15:12:41] <LnxTx> what processor?[15:12:57] <Shadow_aok> Quad-Core AMD Opteron(tm) Processor 8356[15:12:58] *** xinkeT has joined #smartos[15:13:03] <Shadow_aok> "CPU Virtualization": "svm"[15:13:16] *** yruss972 has quit IRC[15:13:16] <Shadow_aok> motherboard is a supermicro H8QM8[15:14:02] <jperkin> you have the AMD patches and are running a custom build?[15:14:08] <yo61> Are AMD processors supported for kvm?[15:14:20] <Shadow_aok> what amd patches ? I only use a standard build.[15:14:24] <yo61> https://github.com/jclulow/illumos-kvm/[15:14:41] <Triskelios> yo61: only in that experimental branch of illumos KVM[15:14:44] <yo61> http://wiki.smartos.org/display/DOC/Hardware+Requirements#HardwareRequirements-KVMRequirements[15:14:51] *** Teknix has quit IRC[15:15:12] <Shadow_aok> damn[15:15:36] <Shadow_aok> so, i need to make a custom build or is there some available ?[15:16:48] *** nikolam has quit IRC[15:17:29] <wiedi> http://imgapi.uqcloud.net/builds <- those have the amd patches[15:18:09] <Shadow_aok> well, it says i need vmx support[15:18:14] <Shadow_aok> to use the mad patches[15:18:15] <Shadow_aok> amd[15:18:31] *** jim80net has quit IRC[15:18:35] <Shadow_aok> don't have it so i guess it's over[15:18:44] *** W0rmDr1nk has joined #smartos[15:18:45] <Shadow_aok> thanks[15:19:38] <Triskelios> Shadow_aok: what? vmx is the Intel thing[15:22:14] <Shadow_aok> i read that on the github page[15:22:21] <Shadow_aok> but ok, it's a relief[15:22:35] <Shadow_aok> so, smartos is already installed and i'm booting from a cd[15:22:48] <Shadow_aok> technically, booting on another cd with smartos and amd support should be enough ?[15:23:19] *** jsm has joined #smartos[15:23:25] <Triskelios> yeah[15:23:50] <Shadow_aok> nice[15:24:00] <Shadow_aok> ok, i'll burn and test it[15:24:04] <Shadow_aok> thanks for the help[15:32:51] *** alcir has quit IRC[15:33:27] <Shadow_aok> anyway, jperkin, you were saying that without AMD support vm creation should still work ? In which case i'm still having an issue with my json file.[15:34:00] *** Xenith_ is now known as Xenith[15:34:46] *** hadret has quit IRC[15:36:16] *** W0rmDr1nk has quit IRC[15:37:03] *** jim80net has joined #smartos[15:37:45] <jperkin> Shadow_aok: in theory yes, though you will probably need to significantly increase the timeout values[15:38:00] <aortmann> hi, my smartos server crashed cause of an power blackout. When I boot now the machoine I got the error "reading beyond end of ramdisk" anyone got an Idea how to fix it? I just have IPMI to access the server[15:38:16] <jperkin> there's little to no merit in actually running without kvm support, though[15:39:06] <Shadow_aok> ok, so that's where the timeout error comes from[15:40:16] <Triskelios> aortmann: what is the boot medium?[15:41:19] <aortmann> it boots from an usb stick[15:42:56] *** W0rmDr1nk has joined #smartos[15:43:06] <Triskelios> aortmann: you may have to reflash it. it's corrupt or something is horribly wrong with your board[15:44:30] <aortmann> mhhh but this wont be easy without coming to the usb stick :D[15:45:20] <MerlinDMC> mount an iso via ipmi and reflash the usb stick after booting from the iso ;)[15:46:00] *** mamash has joined #smartos[15:46:21] <aortmann> this will be my way :)[15:46:23] <aortmann> thanks[15:52:37] <Shadow_aok> ok, booted on the last eait available build.[15:53:30] *** potatosalad has joined #smartos[15:54:44] <Shadow_aok> i have a /dev/kvm now \o/[15:56:34] <Shadow_aok> vm creation is working[15:56:36] <Shadow_aok> thanks all[15:59:18] *** khushildep has quit IRC[16:00:58] *** pringlescan has joined #smartos[16:02:19] *** lumb has quit IRC[16:06:49] *** khushildep has joined #smartos[16:07:21] *** nikolam has joined #smartos[16:09:54] *** tbatchelli_m has joined #smartos[16:11:51] *** lumb has joined #smartos[16:13:01] *** khushildep_ has joined #smartos[16:13:47] *** pringlescan has quit IRC[16:14:30] *** khushildep has quit IRC[16:14:30] *** khushildep_ is now known as khushildep[16:14:51] *** pringlescan has joined #smartos[16:17:09] *** thorfinn has quit IRC[16:17:10] *** pringlescan has quit IRC[16:18:28] *** W0rmDr1nk has quit IRC[16:19:10] *** khushildep has quit IRC[16:21:26] <nahamu> konobi: looks like arekinath has an ipv6 patch too[16:25:27] *** W0rmDr1nk has joined #smartos[16:32:41] *** nefilim1 has joined #smartos[16:33:58] *** mary5030 has joined #smartos[16:34:42] *** nefilim1 is now known as nefilim[16:35:28] *** neophenix has joined #smartos[16:35:53] *** tbatchelli_m has quit IRC[16:37:36] <aortmann> I started the server with a new smartos usb imjage. the local zones all started but the kvm are stopped. Why I have to start the kvm manually?[16:41:17] <aortmann> autoboot is set to true[16:41:19] <aortmann> so :x[16:44:45] <MerlinDMC> aortmann, autoboot gets set to true if you manually start the vm[16:45:04] <MerlinDMC> but it should automatically start after a reboot[16:53:21] <yo61> aortmann: what image?[16:53:36] <yo61> There was a bug that had those symptoms[16:55:29] *** nikolam has quit IRC[16:56:44] *** julianduque has joined #smartos[17:01:42] *** tbatchelli_m has joined #smartos[17:07:06] *** kevinykchan has joined #smartos[17:08:11] *** wiedi has quit IRC[17:09:57] <aortmann> smartos-20130530T224720Z-USB[17:10:02] <aortmann> is the image right now[17:12:04] *** elijah-mbp has quit IRC[17:15:54] <yo61> That could possibly include the bug[17:16:25] <yo61> Any reason not to use the current image?[17:18:04] <aortmann> it seems like that image got the bug :)[17:20:02] *** khushildep has joined #smartos[17:22:13] *** elijah-mbp has joined #smartos[17:22:51] *** daleg has joined #smartos[17:26:59] *** estibi has quit IRC[17:28:47] *** khushildep has quit IRC[17:39:29] *** wolstena has joined #smartos[17:43:04] *** LnxTx has quit IRC[17:43:40] *** LnxTx has joined #smartos[17:44:11] *** dap has joined #smartos[17:47:02] *** lumb has quit IRC[17:47:17] *** jones_d_ is now known as jones_d[17:52:34] *** Shadow_aok has quit IRC[17:54:37] *** lumb has joined #smartos[17:55:56] *** julianduque has quit IRC[17:59:33] *** tbatchelli_m has quit IRC[17:59:55] *** julianduque has joined #smartos[18:00:54] *** nikolam has joined #smartos[18:02:11] *** tbatchelli_m has joined #smartos[18:02:55] *** djun-bb has joined #smartos[18:04:43] *** julianduque has quit IRC[18:05:00] *** tbatchelli_ has joined #smartos[18:06:53] *** djun-bb has quit IRC[18:07:05] *** djun-bb has joined #smartos[18:07:13] *** tbatchelli_m has quit IRC[18:07:17] *** djun-bb has quit IRC[18:07:19] *** julianduque has joined #smartos[18:09:10] *** blindcoder has joined #smartos[18:09:19] <blindcoder> hello.[18:09:50] <blindcoder> Can someone help me with I/O performance on a smartos instance? I have a KVM which only gets 9.6 MB/s read performance on its device[18:10:29] <griffin__> only that one?[18:10:43] <blindcoder> griffin__: One KVM, the rest are zones, I didn't tryb there yet[18:10:47] <blindcoder> I'll check there, just a sec[18:11:57] <blindcoder> tried in one zone, it gets 94 MB/s[18:12:49] *** aortmann has quit IRC[18:13:06] <griffin__> is the kvm using the virtio drivers?[18:13:30] <blindcoder> yes, I have the vmadm get <uuid> output here: http://sprunge.us/AeDf[18:13:38] <blindcoder> "model": "virtio",[18:14:14] <rmustacc> Do you have a slog?[18:14:21] <wesolows_> he's reading.[18:14:21] <rmustacc> What are the nature of the writes of the guest?[18:14:29] <rmustacc> Oh, misread.[18:14:39] <rmustacc> Then I would ask what is the read model?[18:14:42] <wesolows_> That said, it would be useful to see what kind of host I/O the instance is doing.[18:14:44] <rmustacc> Random, streaming, etc?[18:15:00] <blindcoder> I tried dd if=/dev/vdb of=/dev/null bs=1M count=1024[18:15:27] <blindcoder> also, I was doing some imagemagick stuff, creating small tiles from big 25600x25600 png files[18:15:30] <wesolows_> ok, so streaming-only read.[18:15:40] <blindcoder> which shot my load into the 80s[18:16:00] <wesolows_> fwiw, you can do imagemagick in a zone; no need for KVM there.[18:16:17] <wesolows_> I'd follow the usual debugging path for I/O performance here.[18:16:35] <wesolows_> Something like Brendan's USE method is a good place to start, both in the guest and the host.[18:17:00] <blindcoder> http://dtrace.org/blogs/brendan/2012/02/29/the-use-method/[18:17:02] <blindcoder> that one?[18:17:04] <wesolows_> yes[18:17:19] <blindcoder> okay, I'll have a look at it. thanks![18:17:32] <wesolows_> The use of HVM makes it somewhat more complicated, but the basic principles are sound.[18:17:51] <wesolows_> Your biggest problem will likely be inferior tooling in your guest.[18:18:54] <blindcoder> I'm using a pretty standard debian squeeze image[18:21:14] *** julianduque has quit IRC[18:22:13] *** wesolows_ is now known as wesolows[18:27:28] *** altf2o has quit IRC[18:32:16] *** izaki has joined #smartos[18:32:30] *** Alek has joined #smartos[18:33:21] *** ryancnelson has joined #smartos[18:35:04] *** wolstena has quit IRC[18:52:28] *** estibi has joined #smartos[18:53:46] *** _joes_ has quit IRC[18:54:06] *** izaki_ has joined #smartos[18:56:10] *** izaki has quit IRC[18:58:21] <blindcoder> hrm, I'm not sure what to make of this. when reading the same area from the disk using dd several times, the speed goes up like on a cached file[18:59:05] <Licenser> blindcoder if it's ZFS then cache is block based not file based so it makes sense[18:59:21] *** estibi has quit IRC[18:59:32] <blindcoder> Licenser: okay, that makes sense then.[19:00:07] <Licenser> that's why ZFS rocks ;)[19:00:31] <Licenser> that and so many other reasons[19:00:41] <blindcoder> Licenser: yeah, it does, but I still have a linux kvm running which makes this quite unsatisfactory: http://sprunge.us/Wdaa[19:01:14] <ryancnelson> your kvm is still on zfs. you're reading from DRAM the 2nd time[19:01:22] <Licenser> what's bad about this?[19:01:26] <Licenser> what ryancnelson said :)[19:01:40] *** wolstena has joined #smartos[19:01:46] <blindcoder> ryancnelson: yeah, I understand that now :-)[19:02:02] <ryancnelson> it's why windows is actually faster sometimes in our kvm. it's like you replaced ide disks with ramdisks, under-the-hood[19:02:08] <ryancnelson> sometimes[19:02:09] <blindcoder> Licenser: it's bad because a read performance of 10-20 MB/s is skyrocketing my load.[19:02:12] <ryancnelson> depending on workload[19:02:38] <Licenser> wait you're saying it 'reads too fast'?[19:02:46] <blindcoder> Licenser: too slow. 18 MB/s[19:03:14] <blindcoder> Licenser: running something as simple as cp file otherfile can shoot my load into the 80s[19:03:15] *** elijah-mbp has quit IRC[19:03:26] <Licenser> hmmm[19:03:46] <Licenser> perhaps your VM has a very low zfs io priority[19:03:55] *** estibi has joined #smartos[19:04:24] <blindcoder> Licenser: 1024[19:04:33] <blindcoder> Licenser: compared to 32/64/100 on the other zones[19:04:41] <Licenser> just use a zone :P[19:04:45] *** elijah-mbp has joined #smartos[19:04:49] <blindcoder> Licenser: can't run linux in a zone :)[19:05:00] <Licenser> why would you want to run linux ;)[19:05:02] <ryancnelson> that's another reason to use a zone.[19:05:30] <blindcoder> :P[19:06:23] <ryancnelson> "high load" isn't actually a problem. it's just a measurement.[19:06:57] <blindcoder> ryancnelson: yeah, but it causes the system to be unresponsive. apache not serving websites, ssh taking over 2 minutes to log in, etc.[19:06:58] <ryancnelson> "load" == "number of things waiting to get on cpu"[19:07:12] <ryancnelson> that's not because your disks are going too fast, then[19:07:32] <blindcoder> If I run the same task on my laptop on a USB disk, the system remains as responsive as ever[19:07:44] <ryancnelson> that's hardly scientific[19:07:55] <blindcoder> true, but it still confuses the hell out of me[19:07:58] <ryancnelson> you're running in a hypervisor. there's a million things different[19:08:18] <griffin__> what does iostat say?[19:08:46] <ryancnelson> but hey… dial down your arc, if you want to prove/disprove that theory[19:09:01] <ryancnelson> (he wants iostat to be *less* good, griffin__ )[19:09:06] <blindcoder> griffin__: iostat in the guest goes to roughly 5000 kB/s read performance with cache misses[19:09:13] *** vizant0 has quit IRC[19:09:31] <griffin__> what about wait and busy?[19:09:33] <blindcoder> ryancnelson: actually, I want that task to not cause the system to slow down to a crawl[19:09:46] *** vizant0 has joined #smartos[19:10:05] *** vizant0 has quit IRC[19:10:25] <blindcoder> griffin__: let me start the task[19:11:33] <blindcoder> griffin__: currently at 80%[19:11:55] <blindcoder> but I see now that that's while writing, not reading[19:12:13] <ryancnelson> are you using virtio, or ide storage drivers in your vm?[19:12:18] <blindcoder> virtio[19:13:09] <griffin__> this task uses a lot of proccesses?[19:13:22] <blindcoder> griffin__: no, just one call to convert from imagemagick[19:13:57] <blindcoder> this is it: convert current-0-0.png -crop 512x512 -set filename:tile %[fx:page.x/512+0]_%[fx:page.y/512+0] map_files/19/%[filename:tile].jpg[19:14:15] <griffin__> but just a simple dd if=/dev/zero of=file is slow as well?[19:14:46] <blindcoder> griffin__: I'll try[19:14:58] <blindcoder> just waiting for the io to die down...[19:15:48] <griffin__> you could try this as well to enable zfs prefetch: echo zfs_prefetch_disable/W0t0 | /usr/bin/mdb -kw[19:16:59] <blindcoder> iowait also goes to 50-60%, but dd returns immediately[19:17:19] *** ryancnelson has left #smartos[19:17:28] <blindcoder> it might not be the disks after all...[19:17:52] <griffin__> the convert just slow?[19:18:05] <blindcoder> yeah...[19:22:00] *** pringlescan has joined #smartos[19:25:26] <blindcoder> it's probably just really much i/o. I totally forgot that convert needs about 12 GB of memory for png files of this size[19:26:09] <blindcoder> and if it doesn't have enough RAM, or it's RAM usage has been restricted, it resorts to creating temporary files[19:26:28] <nahamu> *cough* manta *cough*[19:26:42] <blindcoder> manta?[19:27:24] <nahamu> http://www.joyent.com/products/manta (also relevant: http://www.joyent.com/blog/manta-image-content-manipulation-and-publishing-example-part-1 )[19:28:41] <pringlescan> does anyone else have BSOD on Windows 2008? they're always BAD_POOL_HEADER in mountmgr.sys[19:28:50] <nahamu> but that's a tangent if you really need to do this stuff locally on your SmartOS box.[19:30:33] <blindcoder> nahamu: I'm not quite sure if I understand what you're trying to say.[19:30:56] <blindcoder> nahamu: I have 66 PNG files here, totalling at over 8 GB in size and being a total of 42.25 gigapixels[19:31:20] <blindcoder> nahamu: I'm creating the deepzoom tiles for an openseadragon instance here.[19:31:22] *** bens1 has quit IRC[19:32:53] <nahamu> That might be a perfect candidate for uploading the originals into Joyent's manta service, and doing all the processing there, and possibly even hosting the results there as well (I've never used openseadragon)[19:33:20] <nahamu> But if you need to do everything on your local machine for whatever reason, then you should ignore me.[19:33:47] <blindcoder> nahamu: I don't necessarily need to, I've just been doing it so far :)[19:36:16] <nahamu> right. I recommend reading the blog post I linked, and there's #manta if you have questions about manta in general[19:38:05] <blindcoder> nahamu: thanks :)[19:38:14] <nahamu> but this suggestion of mine is a bit off-topic for #smartos and perfect for #manta. :)[19:43:41] *** julianduque has joined #smartos[19:43:48] *** wolstena has quit IRC[19:48:01] <konobi> arekinath: about?[19:48:01] *** bens1 has joined #smartos[19:48:15] *** tbatchelli_ has quit IRC[19:48:54] *** bens1 has quit IRC[19:56:25] *** tbatchelli_m has joined #smartos[20:03:20] *** noahmehl has joined #smartos[20:05:54] *** bens1 has joined #smartos[20:08:40] *** leecallen35 has joined #smartos[20:08:46] *** tbatchelli_ has joined #smartos[20:09:35] *** leecallen has quit IRC[20:11:12] *** wolstena has joined #smartos[20:11:58] *** tbatchelli_m has quit IRC[20:13:49] *** bens1 has quit IRC[20:27:15] *** tbatchelli_ has quit IRC[20:28:22] *** pringlescan has quit IRC[20:29:32] *** pringlescan has joined #smartos[20:39:04] *** paul_lamb has joined #smartos[20:44:45] *** jaimef_ has quit IRC[20:45:05] *** mamash has left #smartos[20:47:14] *** jaimef has joined #smartos[20:50:33] *** notklaatu has joined #smartos[21:23:15] *** ryancnelson has joined #smartos[21:25:05] <pettson> Hm, pkgin in ruby193 installs PHP. dafaque?[21:25:39] <ryancnelson> which pkgin repo are you using?[21:27:05] <pettson> http://pkgsrc.joyent.com/packages/SmartOS/bootstrap/bootstrap-2013Q2-x86_64.tar.gz ?[21:27:50] <MerlinDMC> pettson, if you did switch to that repo from an earlier one you should maybe rebuild the package database[21:29:27] <pettson> MerlinDMC: Newly created VM, then follow this http://wiki.smartos.org/display/DOC/Installing+pkgin with the above mentioned URL.[21:30:25] <ryancnelson> you did it wrong, then[21:30:34] <ryancnelson> you dont' need to install pkgin *inside* a vm[21:30:43] <ryancnelson> it's already there[21:30:49] *** chorrell has joined #smartos[21:30:53] <ryancnelson> those instructions are for putting pkgin into the global zone[21:31:33] <ryancnelson> so, you stomped on your pkgin installation[21:33:03] *** mamash has joined #smartos[21:34:07] <ryancnelson> I just updated the wiki page to make that clear.[21:34:20] <pettson> ryancnelson: Thanks.[21:34:21] <pettson> cool[21:35:00] <pettson> I'm a bit confused. I keep doing one thing like once a month on my smartos-box. :)[21:36:13] <pettson> Luckily it's easy to rebuld the VM.[21:36:31] <pettson> Now the pkgin in dependencies looks nicer.[21:36:34] <ryancnelson> snapshot/rollback is my buddy[21:39:45] <pettson> Is there a way to upgrade smartos-vms?[21:41:17] <pettson> Like, if i base them on base64 13.1.0 smartos, and want to upgrade to 13.2.0?[21:41:28] <MerlinDMC> you could switch to a newer pkgin repository and reinstall everything in /opt ... but that would not catch everything that was probably done to the configuration in /etc[21:42:45] <MerlinDMC> I always recreate and reprovision the zones - web projects are stored on some filesystem mount and can be remounted in the new zone so that I don't need to care about changes there much[21:43:24] <MerlinDMC> chef/puppet your way[21:44:49] <pettson> ok[21:45:12] <pettson> And the same if I would upgrade the global zone (http://wiki.smartos.org/display/DOC/Remotely+Upgrading+A+USB+Key+Based+Deployment) ?[21:45:33] <MerlinDMC> upgrading the GZ ist just a update your boot media and reboot[21:45:55] <pettson> So the VMs don't use anything from the GZ?[21:46:41] <MerlinDMC> some folders get mounted ... /lib and /usr so there're some tools that get used from the GZ but your zones will (should) work like before[21:47:13] <pettson> "should" :)[21:47:23] <MerlinDMC> at least if you don't focus your apps to rely on stuff in /lib or /usr ... which you shouldn't[21:47:43] <MerlinDMC> prefer stuff you install via pkgin inside /opt and you should be fine[21:48:11] <MerlinDMC> "should" just because you can break almost everything if you want to[21:48:31] <pettson> Right, as long as there are no strange ABI-incompatibilities between the VM-version and GZ.[21:49:07] <MerlinDMC> and even if ... downgrading is not a problem if you write your older version to your boot media and just reboot[21:54:25] <pettson> Is there some mailing list for security issues btw?[21:54:43] *** Alek has quit IRC[21:57:18] *** leecallen has joined #smartos[21:59:24] *** Alek has joined #smartos[22:01:18] *** leecallen35 has quit IRC[22:05:57] <ryancnelson> there will be no abi incompatibilities between the zone and the gz[22:06:19] <ryancnelson> the zone is basically nothing except software you've installed[22:06:24] *** jaimef has quit IRC[22:07:06] <ryancnelson> that's precisely how it was designed to work. all zones are containers, sharing one kernel from the GZ[22:08:18] <pettson> ryancnelson: Yes, and historically I think solaris have a good record of keeping abi compat between kernel and user land.[22:08:51] <pettson> But that could easily change. :-9[22:08:52] <rmustacc> pettson: We ensure that the only things exposed by the platform and have compilation symlinks in todays smartos are interfaces that we have committed to and well ensure forwards compatability.[22:08:55] *** W0rmDr1nk has quit IRC[22:09:05] <pettson> cool[22:09:13] <ryancnelson> put another way: joyent's entire public cloud business it dependant on that not changing.[22:09:35] <ryancnelson> otherwise, we can't update our kernels without breaking customers.[22:09:42] <pettson> aight[22:09:45] <ryancnelson> it's not a happy side-effect. it's goal number 1[22:10:43] *** jaimef has joined #smartos[22:10:58] <pettson> Speaking about happy side-effects.. how much of a "happy side-effect" of SDC is SmartOS as an operating system.[22:11:52] <ryancnelson> that's not a complete sentence[22:11:54] <ryancnelson> what?[22:12:04] <pettson> ryancnelson: I mean, I fully understand that SmartOS being able to run in your SDC-setup is goal no.1. But on some kind of scale - how important is it to Joyent that it is usable by other people as well?[22:12:42] *** vizant0 has joined #smartos[22:13:18] <rmustacc> pettson: We have rather specific design choices that we've made. If you fit into those, great.[22:13:25] <wesolows> the non-product on smartos.org is a happy side effect.[22:13:26] <rmustacc> If not, we generally aren't changing.[22:13:51] <wesolows> About 98% of it's shared with SDC's platform anyway, so it's very easy to make available for people to use and share and play with.[22:14:20] <wesolows> We also sell SDC, so it's not used only in our own data centres.[22:14:23] <ryancnelson> it's not really a side-effect, though. we release it because we're proud of it, not because we're satisfying some legal license clause[22:14:26] <griffin__> that was one of the most honest answers[22:14:29] <wesolows> It's a multipurpose platform.[22:14:42] <ryancnelson> we package it into a usable OS[22:15:03] <ryancnelson> not just leak some tarball of source every 12 months[22:15:14] <ryancnelson> we do that every two weeks[22:15:15] <wesolows> Yes, definitely. If we just wanted to satisfy license requirements we'd do what Sun used to do and give people a DVD with a bunch of source code on it.[22:15:18] <wesolows> That'd be stupid.[22:15:58] <ryancnelson> smartos, as released is almost all of the OS portion of SDC.[22:16:23] <ryancnelson> it's just the orchestration parts that aren't included in smartos.org[22:16:43] <pettson> I hope you continue being proud of it then.[22:16:53] <ryancnelson> ... which i realize is what wesolows said 2 minutes ago[22:17:42] <ryancnelson> typically, you wouldn't accept pull requests on a happy side effect :)[22:18:10] <nahamu> I think Joyent does benefit from the community of people who sometimes trip over new bugs before they can make it into SDC.[22:18:16] <wesolows> I'm happy to have people fix things. We just want to be clear that we're setting the technical direction.[22:18:42] <wesolows> Absolutely.[22:19:07] <wesolows> The world gets a good small-scale hypervisor and we get exposure and the occasional fix.[22:19:25] <nahamu> small-scale?[22:19:32] <pettson> Would you accept pull requests that aren't at all useful for you, but could be useful for others? (as long as it technically makes sense, and isn't directly harmful)[22:19:37] <ryancnelson> nahamu: running a public cloud not only pays the bills, it's the ultimate shakedown test[22:19:40] <wesolows> As delivered, SmartOS is really only useful on a single system.[22:19:55] <nahamu> Ah, I see what you mean.[22:20:08] <wesolows> pettson: Those qualifications are important.[22:20:15] <nahamu> ryancnelson: indeed. I am similarly happy to be running an OS that can withstand powering the JPC.[22:20:17] <ryancnelson> think about it: people say "i'm not running that code on my production systems... test it in the cloud"[22:21:05] <pettson> wesolows: Of course. I'm just curious whether you'd refuse stuff that isn't a direct benefit for you.[22:21:07] <ryancnelson> ... that, plus the dicks that buy vm's to attack other people. our "multi-tenant" features had better be robust, or we'd be hosed[22:21:18] <wesolows> Usually not, but it depends on the details.[22:21:31] <wesolows> (i.e., usually we won't reject stuff just because we don't care)[22:21:51] <wesolows> There are exceptions, though; something that's invasive or high risk or would be a support or maintenance burden won't be accepted.[22:22:00] <nahamu> pettson: if you can put your stuff into an overlay, you could spin your own builds and stay current with exremely minimal effort.[22:22:12] <nahamu> *extremely[22:22:13] <wesolows> And yeah, it's very easy to fork if you want to go your own way.[22:22:18] <ryancnelson> we might refuse stuff that's frivolous, though. no soundblaster drivers, please[22:22:20] <wesolows> We encourage that, actually.[22:22:26] <pettson> nahamu: But I have no interest in running my own unix distribution. :)[22:22:34] <nahamu> pettson: what features do you have in mind?[22:22:46] <wesolows> Also, if you're working on the OS specifically, everything should go through illumos, which we don't control.[22:22:47] <MerlinDMC> pettson, there were some pull-requests in the past for features not really used on SDC afaik ... spice protocol port and some additions to the VNC configurations[22:23:24] <MerlinDMC> just as two little examples[22:23:58] <nahamu> yeah, spice might be the poster child for a feature completely ignored by SDC[22:24:41] <pettson> nahamu: Nothing really in particular (maybe config options for lofs in vmadm). But sometimes when reading the interwebs it's not too uncomon to read things like "do you suport $foo?" "No. $foo is not needed in SDC."[22:25:04] <wesolows> config options for lofs seem fine.[22:25:06] <griffin__> yeah my sb16 isnt working[22:25:37] <pettson> wesolows, ryancnelson: I'm sorry if I give an impression of being ungrateful or something. I think SmartOS is a really cool product.[22:25:51] <pettson> you rock :)[22:25:54] <paul_lamb> can i attached a physical drive to a linux vm instance?[22:25:55] <wesolows> These are perfectly good questions to be asking.[22:25:57] *** chorrell has quit IRC[22:26:28] *** estibi has quit IRC[22:26:45] <ryancnelson> paul_lamb: there's not device passthru[22:26:45] <pettson> wesolows: I guess I got burnt by the Oracle-buying-Sun-story. :)[22:26:57] <wesolows> You're certainly not the only one.[22:26:59] <ryancnelson> so did he, pettson[22:27:00] <ryancnelson> :)[22:28:00] <pettson> Yeah, I guess it was way worse for others. I'm just using solaris for my personal stuff.[22:28:31] <nahamu> as I think others have pointed out, not only does Joyent release the latest build every 2 weeks and publish all the source, they work incredibly hard to make it as easy as possible for anyone to get a build environment up and running and replicate the build themselves.[22:28:31] <griffin__> yeah it could be worse, you could for instance be locked into huge support contracts on thousands of servers[22:28:45] <griffin__> and don't think you'll get any synergy from also using oracle databases[22:29:22] <pettson> griffin__: Then it wouldn't be personal. =)[22:29:24] <wesolows> The other thing we work on, when we have time, is elimination of the few remaining dependencies on closed-source stuff.[22:29:31] <wesolows> (none of which is ours anyway)[22:31:04] <griffin__> you have any idea which 'competitor' of yours is funding all the vps/hypervisor stuff in freebsd?[22:31:21] <wesolows> I don't, anyway.[22:31:29] <ryancnelson> by all means, use the existence of smartos as a iron bar to bash your sales reps from other big companies with.[22:31:30] <wesolows> Nor was I really aware someone was.[22:31:38] <ryancnelson> hypervisor in freebsd?[22:32:00] <griffin__> yes[22:32:03] <ryancnelson> not the beehive stuff?[22:32:09] <griffin__> yeah bhyve[22:32:15] <ryancnelson> sorry: "BHyVe"[22:33:02] <ryancnelson> that only supports booting freebsd on freebsd, right?[22:33:31] <ryancnelson> not a huge threat, business-wise. good excersise for the virtio drivers, though. I'd thank them for that[22:35:10] <MerlinDMC> ryancnelson, what is the "huge thread" currently? ;)[22:35:16] <griffin__> I dont know where its at with the bios emulation, but they've booted centos on it, so I dont think its far away from running other guests[22:35:16] <nahamu> "CentOS under bhyve was demonstrated at the BSDCan 2013 FreeBSD Developer Summit."[22:35:41] *** bens1 has joined #smartos[22:36:02] <nahamu> eh, it's BSD licensed... if it's really good, Joyent can just switch from KVM to bhyve.[22:36:16] <nahamu> hell, illumos could actually pull that stuff in if they wanted.[22:38:54] *** bens1 has quit IRC[22:38:55] <griffin__> yeah the 'competitor' isnt the core OS, its whoever is going to lay the SDC type stuff over it[22:41:13] <ryancnelson> i am *not* a biz-dev guy, but i'm talking about what people want to run as a "vps". it's not usually freebsd.[22:41:18] <ryancnelson> we offer a freebsd vm, though.[22:41:20] <ryancnelson> i like it[22:41:27] <ryancnelson> ryan.net is freebsd, since forever[22:41:56] <ryancnelson> hypervisors are a concession, though, to people who need it to run their thing[22:43:26] <ryancnelson> (or, for people who're willing to pay the performance hit, to be able to forklift their exact environment into a vm)[22:46:03] *** bixu has joined #smartos[22:52:11] *** FatDarrel has joined #smartos[22:52:58] <konobi> ryancnelson: freebsd are working on a kvm port based on our one =0)[22:53:49] <ryancnelson> they had one, long ago. did they toss it out?[23:00:39] <konobi> not sure[23:13:57] *** mgls has joined #smartos[23:16:42] *** izaki_ has quit IRC[23:23:20] <griffin__> im pretty sure the kvm is tossed in favor of bhyve[23:23:30] *** Alek has quit IRC[23:31:30] *** mamash has left #smartos[23:33:12] <jperkin> pettson: going back to an earlier question about security mailing list, no, but there is "pkg_admin audit" which will tell you all the vulnerabilities (assuming your vuln file is up-to-date) with your currently installed packages[23:33:26] *** ryancnelson has left #smartos[23:34:02] <jperkin> the default cron fetches the vulnerabilities file, so you shouldn't need to do it manually, but if you want to - run pkg_admin fetch-pkg-vulnerabilities[23:34:41] <bixu> jperkin: But the cron doesn't install patches, correct? Just updates the local index or something like that?[23:36:38] <jperkin> right[23:37:07] <pettson> jperkin: And if there would be a security issue with SmartOS itself? :)[23:38:22] <jperkin> pettson: then it'd be rolled into the next bi-weekly update (but I'm sure we could do a faster update if necessary)[23:38:50] *** wiedi has joined #smartos[23:41:25] <pettson> jperkin: But I don't want to reboot my server every second week just to maybe get patches I'm interested in.[23:43:00] <rmustacc> Well, if it's a bug in the platform that's your only option.[23:43:13] <rmustacc> If it's not in the platform and one of the many pieces of third party software in pkgsrc then there's nothing to do.[23:44:22] <rmustacc> To a certain extent with most things in the platform you cannot safely upgrade it without taking a reboot due to needing the kernel and its core libraries being in sync.[23:46:47] <pettson> rmustacc: But the suggestion was to upgrade every second week, regardless of the nature of the upgrade.[23:47:21] <jperkin> you don't _need_ to upgrade every two weeks, but if you want to get the latest stuff, then yes you need to reboot[23:47:54] <jperkin> unless there is specific new functionality in vmadm or something, then you might be able to hack some lofs mounts..[23:48:10] *** paul_lamb has quit IRC[23:48:25] <pettson> jperkin: But the context was security fixes now.[23:49:10] <jperkin> sorry, I'm struggling to see your point[23:49:17] <pettson> I may not want to upgrade my installation every second week. But if there is some security issue I want to.[23:49:21] <jperkin> if there's a security fix, upgrade and reboot.[23:49:35] <pettson> jperkin: No shit.[23:50:01] <pettson> jperkin: But how would I get the information about the security fix?[23:50:07] <jperkin> from the changelog[23:50:25] <rmustacc> We try to send notices about those as we encounter them to the discuss list.[23:50:46] <jperkin> and if you want more specifics you can look at the git history and show the exact code changes[23:50:59] <rmustacc> But the question of what is security or what is not is not always 100% clear I guess.[23:51:35] *** neophenix has quit IRC[23:52:00] <pettson> jperkin: And the changelog is sent to the discuss-list?[23:52:31] <jperkin> it's listed in every release announcement, yeh.[23:52:34] <wesolows> It's important to understand that the number of significant security problems in the platform tends to be fairly small.[23:52:56] <wesolows> The platform itself is small. So most biweekly builds don't contain new fixes for security problems.[23:53:33] <wesolows> Therefore, if your sole criterion for rebooting (i.e., upgrading the platform) is to pick up fixes for serious security-related bugs, you'll probably upgrade only a few times a year.[23:53:35] <pettson> Of course. But _if_ some nasty shit turns up I'd sort of like to know. =)[23:53:37] <wesolows> Not every 2 weeks.