Switch to DuckDuckGo Search
   March 4, 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 | 31 | >

bottom
[00:03:34] *** tonyarkles has quit IRC
[00:06:15] *** avrntsv has quit IRC
[00:14:28] *** colstrom has quit IRC
[00:20:02] *** denizr has quit IRC
[00:32:13] <rmustacc> alonsodev: If you're trying to create an Ubuntu VM that's the wrong page.
[00:32:15] *** szaydel has joined #smartos
[00:32:40] <rmustacc> *alfonsodev: ^
[00:39:39] *** alfonsodev has quit IRC
[00:39:57] *** tonyarkles has joined #smartos
[01:36:32] *** seancribbs has joined #smartos
[01:43:47] *** leecallen has quit IRC
[01:43:50] *** leecallen35 has joined #smartos
[01:52:16] *** seancribbs has quit IRC
[02:05:49] *** szaydel has quit IRC
[02:20:24] *** tonyarkles has quit IRC
[02:41:43] *** andywocky has joined #smartos
[02:45:11] *** dizko has joined #smartos
[02:51:35] *** denizr has joined #smartos
[02:57:18] *** szaydel has joined #smartos
[03:02:25] *** tonyarkles has joined #smartos
[03:11:28] *** enmand has joined #smartos
[03:13:33] *** krisa has joined #smartos
[03:33:28] <ChrisPartridge> What versions of KVM/QEMU is the latest smartos using?
[03:34:35] *** szaydel has quit IRC
[03:35:27] *** szaydel has joined #smartos
[03:35:33] *** Cpt-Oblivious has quit IRC
[03:39:50] <nahamu> ChrisPartridge: https://github.com/joyent/illumos-kvm-cmd
[03:40:26] *** des2 has quit IRC
[03:40:30] <ChrisPartridge> ty
[03:40:31] <nahamu> Their custon patched version originally forked somewhere around qemu-kvm 0.14
[03:40:34] <nahamu> *custom
[03:41:04] <nahamu> If you want to use something based on qemu-kvm-1.1.2, I have a build that can be used as an add-on that also includes spice support.
[03:41:25] <nahamu> http://wiki.smartos.org/display/DOC/Spice+on+SmartOS
[03:43:46] <nahamu> </shameless plug>
[03:44:17] <ChrisPartridge> cool, I'll take a look
[03:44:39] <ChrisPartridge> was more interested to see if any work was done on qemu 1.4 (they note a heap of performance increases)
[03:45:39] *** trentster has quit IRC
[03:46:41] *** enmand_ has joined #smartos
[03:47:11] *** des2 has joined #smartos
[03:47:20] <nahamu> ChrisPartridge: Since my time to hack on things is limited, I'm always looking for collaborators. If you have any interest in trying to use my tree as inspiration for patching 1.4 to work on SmartOS, it would be very welcome. :)
[03:47:43] *** enmand_ has quit IRC
[03:48:04] <nahamu> Ultimately I'd love to get the right patches accepted upstream so that at least for my project I could just be a downstream consumer rather than maintaining a fork.
[03:48:08] <dizko> I'm interested in sharing data between an os vm and a kvm, seems nfs is probably the only way.  Channel logs indicate someone doing that was getting 30% cpu in system time while transferring
[03:48:22] <dizko> anyone know if that's been addressed / is accurate?
[03:49:12] *** trentster has joined #smartos
[03:49:16] *** enmand has quit IRC
[03:49:21] <nahamu> I would imagine the Joyent folks might also be interested but just don't have the time at the moment.
[03:49:53] <nahamu> what kind of performance increases are we talking about?
[03:50:13] <dizko> http://echelog.com/logs/browse/smartos/1360537200
[03:50:23] <dizko> nefilim mentioning the bad performance with that setup
[03:50:39] <ChrisPartridge> nahamu: it's marked as experimental, but according to the release "experimental threaded backend for virtio-blk-pci achieving up to a 900% increase in IOP rate on very large storage devices"
[03:51:24] <nahamu> That's very interesting. Possibly even interesting enough to get the attention of the Joyent people. I'm glad you mentioned it.
[03:51:32] <Zigara> that is indeed very interesting
[03:51:50] <ChrisPartridge> nahamu: Looking here http://dtrace.org/blogs/brendan/2013/01/11/virtualization-performance-zones-kvm-xen/ at brendans blog, the IO performance kinda sucks in comparison to zone
[03:52:04] *** enmand has joined #smartos
[03:52:18] *** enmand has quit IRC
[03:53:00] <nahamu> Zones should always win over KVM VMs, but closing the performance gap would just further Joyent's ability to tout their cloud having the best performance.
[03:55:05] *** avrntsv has joined #smartos
[03:56:11] <nahamu> rmustacc: see above when you get a chance...
[03:59:04] *** enmand has joined #smartos
[04:03:14] <ChrisPartridge> nahamu: excuse my smartos ignorance, but essentially I can use different builds of qemu, in different zones if I please?
[04:04:10] <nahamu> Yes and no.
[04:04:19] <nahamu> Generally not recommended unless you have a good reason.
[04:04:51] <nahamu> My good reason was spice support. So technically yes, there is a feature where you can provide a ZFS file system with an alternate build of QEMU
[04:05:00] <nahamu> They pretty much added it for me.
[04:05:28] <nahamu> Your QEMU needs to have the relevant Joyent patches applied, so you can't just use upstream QEMU and have it work (at least I don't think you can)
[04:06:40] <nahamu> but if you don't mind getting your hands dirty with code, then yes, you can use different builds of QEMU in different zones.
[04:09:57] <rmustacc> nahamu, ChrisPartridge: Yes, I've seen it perviously. Unfortunately experimental is also a code name for do not use in my book. Furthermore, most of the disk I/O issues that people have are best solved with a slog, since we're not changing the default behavior away from synchronous writes unlike qemu upstream.
[04:09:59] *** trentster has quit IRC
[04:10:46] <rmustacc> For the majority of VMs out there, the block I/O isn't the problem.
[04:11:22] *** trentster has joined #smartos
[04:12:25] <nahamu> rmustacc: interesting.
[04:13:02] <rmustacc> Generally with a bit of ARC + a slog you solve most of the block device problems that you have.
[04:13:11] <nahamu> I agree that while it's still experimental it doesn't belong in production
[04:13:15] <rmustacc> I mean, I'm curious to see what it is exactly that they did, but it also is far from a priority. Especially you know, that whole experimental thing.
[04:13:23] <rmustacc> So I'll reserve looking at it until it actually is solid.
[04:13:57] <nahamu> I think you make an important point about what it's actually doing, and is it solving a problem that SmartOS solves with a SLOG.
[04:14:10] <rmustacc> Well, no, I don't think it does.
[04:14:18] <rmustacc> I don't know what the problem they are trying to solve there is.
[04:14:34] <rmustacc> Even with a multi-threaded virtio for block devices, that doesn't solve the problem of making sure writes are correct.
[04:14:48] <rmustacc> There are still a large number of guest filesystems that don't use barriers at all.
[04:15:00] <rmustacc> So emulating a disk with a cache in qemu is hazardous because of that.
[04:15:01] <nahamu> I would have thought that for speed it's a matter of more threads to drain the I/O ring buffers or something
[04:15:24] <nahamu> but clearly I haven't looked much further than that changelog page
[04:16:17] <rmustacc> Who knows. It's also IOPS.
[04:16:20] <nahamu> at any rate it sounds like your first impression is "solution looking for a problem" when it comes to SmartOS.
[04:16:32] <rmustacc> Not quite.
[04:16:37] <rmustacc> My first impression is that I don't use experimental stuff.
[04:17:34] <rmustacc> Not that updating to a newer qemu isn't something I want to do at some point, but I'm also skeptical of it. Especially generally speaking that most of the bottlenecks people have in our guests is network based.
[04:17:37] <nahamu> So you wait for us to run the latest builds of SmartOS? :-P
[04:17:59] <rmustacc> No, we run them.
[04:19:20] <nahamu> I'd argue that on the day they're released, they're "experimental" too, but I need to go to bed, so I will concede that in this case, a feature in QEMU marked as experimental is a bit more suspect than a SmartOS build.
[04:20:02] *** andywocky has quit IRC
[04:20:03] <nahamu> What I'm getting at is that block IOPS aren't a pain point for SmartOS right now, so that's not the angle to pitch "SmartOS needs the latest QEMU"
[04:20:15] <rmustacc> Well, not for things I look at.
[04:20:18] <nahamu> right
[04:20:24] <rmustacc> It may be for others.
[04:20:44] <rmustacc> And don't forget not all IOPS are created equal.
[04:21:59] <nahamu> Bedtime for me, but thanks for your time, rmustacc. And Thanks ChrisPartridge for pointing out the QEMU 1.4 release notes.
[04:22:22] <nahamu> goodnight!
[04:23:03] *** trentster has quit IRC
[04:23:11] *** avrntsv has quit IRC
[04:23:35] *** trentster has joined #smartos
[04:24:43] *** tonyarkles has quit IRC
[04:34:16] *** trentster has quit IRC
[04:51:11] *** szaydel has quit IRC
[04:53:30] *** trentster has joined #smartos
[05:04:44] *** iyp has quit IRC
[05:07:50] *** Azbruh has quit IRC
[05:09:09] *** Azbruh has joined #smartos
[05:13:46] *** krisa has quit IRC
[05:15:51] *** trentster1 has joined #smartos
[05:16:09] *** trentster has quit IRC
[05:16:15] *** trentster1 has quit IRC
[05:29:02] *** trentster has joined #smartos
[05:31:08] *** szaydel has joined #smartos
[05:31:09] *** tonyarkles has joined #smartos
[05:53:16] *** denizr has quit IRC
[05:56:10] *** andywocky has joined #smartos
[06:10:25] *** enmand has quit IRC
[06:21:01] *** andywocky has quit IRC
[06:43:47] *** szaydel has quit IRC
[07:07:34] *** porkbelt has quit IRC
[07:19:28] *** wolfeidau has quit IRC
[07:20:26] *** tonyarkles has quit IRC
[07:24:38] *** ipalreadytaken has joined #smartos
[07:45:03] *** wolfeidau has joined #smartos
[07:45:50] *** ChrisPartridge has quit IRC
[07:49:38] *** porkbelt has joined #smartos
[07:54:26] *** ipalreadytaken has quit IRC
[07:59:10] *** nefilim has quit IRC
[07:59:44] *** deedubs has quit IRC
[07:59:48] *** Raynos has quit IRC
[08:04:37] *** abhishekvasisht has joined #smartos
[08:08:10] *** bsdguru has quit IRC
[08:23:02] *** alucardX has joined #smartos
[08:28:06] *** sachinsharma has joined #smartos
[08:31:33] *** ipalreadytaken has joined #smartos
[08:38:54] *** arekinath has quit IRC
[08:42:02] *** alucardX has quit IRC
[08:46:15] *** serengeti has joined #smartos
[08:47:36] *** alucardX has joined #smartos
[08:50:07] *** cwvhogue has joined #smartos
[08:51:01] *** cwvhogue has left #smartos
[08:54:29] *** sachinsharma has quit IRC
[08:54:54] *** sachinsharma has joined #smartos
[08:59:14] *** ipalreadytaken has quit IRC
[09:20:24] *** texarcana has quit IRC
[09:21:39] *** texarcana has joined #smartos
[09:22:29] *** alfonsodev has joined #smartos
[09:23:10] *** bens1 has joined #smartos
[09:29:46] *** ipalreadytaken has joined #smartos
[09:32:41] *** KermitTheFragger has joined #smartos
[09:38:46] *** ipalreadytaken has quit IRC
[09:49:56] *** abhishekvasisht has quit IRC
[09:55:59] *** abhishekvasisht has joined #smartos
[10:10:37] *** bluezenix has joined #smartos
[10:12:16] *** serengeti has left #smartos
[10:13:10] *** chh has joined #smartos
[10:27:56] *** deedubs has joined #smartos
[10:28:58] *** Raynos has joined #smartos
[10:33:10] *** jamesd_ has quit IRC
[10:35:17] *** ipalreadytaken has joined #smartos
[10:37:21] *** darjeeling has quit IRC
[10:43:22] *** ipalreadytaken has quit IRC
[10:44:51] <alfonsodev> @rmustacc can you point me to the correct one? or give me a hint for the keywords to search for ?
[10:51:44] <MerlinDMC> alfonsodev, for vms use 'kvm' as brand
[10:52:22] <MerlinDMC> everything other than smartos (everything using qemu/kvm) should use kvm as brand
[10:52:55] *** sachinsharma has quit IRC
[10:53:21] <MerlinDMC> but also the rest of the json is pretty different
[11:06:39] *** sachinsharma has joined #smartos
[11:09:05] *** alfonsodev has quit IRC
[11:21:06] *** sachinsharma has quit IRC
[11:24:18] *** sachinsharma has joined #smartos
[11:24:41] *** abhishekvasisht1 has joined #smartos
[11:26:48] *** abhishekvasisht has quit IRC
[11:45:54] *** alucardX has quit IRC
[11:47:55] *** alucardX has joined #smartos
[12:48:01] *** stetho has joined #smartos
[12:59:14] *** abhishekvasisht1 has quit IRC
[13:03:29] *** abhishekvasisht has joined #smartos
[13:10:52] *** sachinsharma has quit IRC
[13:38:35] *** szaydel has joined #smartos
[13:41:30] *** ipalreadytaken has joined #smartos
[13:43:24] *** leecallen35 has quit IRC
[13:43:54] *** leecallen has joined #smartos
[13:47:51] *** ira has joined #smartos
[13:53:01] *** kamilr has joined #smartos
[14:04:34] *** estibi has quit IRC
[14:21:33] *** enmand has joined #smartos
[14:47:20] *** estibi has joined #smartos
[14:47:32] *** darjeeling has joined #smartos
[14:47:56] *** natefoo_ is now known as natefoo
[14:55:13] *** khushildep has joined #smartos
[14:55:40] *** khushildep has quit IRC
[14:56:37] <KermitTheFragger> hi all
[14:56:55] <KermitTheFragger> has anyone ever gotten further with SmartOS and desktop virtualization? I saw on the wiki there was some interest
[14:57:34] <MerlinDMC> KermitTheFragger, nshalman is using spice to provide some vdi afaik
[14:58:57] <KermitTheFragger> MerlinDMC: yeah i found his wiki page: http://wiki.smartos.org/display/DOC/Spice+on+SmartOS
[15:01:48] <KermitTheFragger> MerlinDMC: but there are no other initiatives that you know of? (easy to use end user interface for example **crosses fingers**)
[15:02:34] <MerlinDMC> none that I know ... sry
[15:05:53] <nahamu> KermitTheFragger: We wrote our own thin client software wrapper around the spice client running on Fedora.
[15:06:28] <nahamu> But we have particular security needs, so our client is probably overkill for most people.
[15:07:17] <nahamu> There are people working on Mac spice clients and a client written in HTML5 served as a web page, but I haven't really used either.
[15:07:55] <nahamu> If VNC provides enough performance for you, I'd try out Project FiFo as it provides VNC access from the web browser.
[15:07:55] <KermitTheFragger> nahamu: out of curiosity; big deployment? does it work ok for you guys (stable, scalable and such)?
[15:08:41] <nahamu> ~50 users, ~2 VMs per user, not perfectly stable, but stable enough that the users aren't calling for my head.
[15:08:58] *** arekinath has joined #smartos
[15:08:59] *** arekinath has joined #smartos
[15:09:15] <nahamu> not sure what you mean by scalable. currently managing it is not ideal
[15:09:15] <KermitTheFragger> what are the users running on their VM's ? *nix or Windows?
[15:09:35] <Zigara> linux!
[15:09:37] <nahamu> mostly Fedora
[15:09:41] <nahamu> a few Windows VMs
[15:09:55] <nahamu> spice performance might actually be better for the Windows VMs than the linux ones though
[15:10:09] <nahamu> Windows was the original target of Spice development, as I understand it.
[15:10:12] <Zigara> meh, I was not super impressed with spice
[15:10:21] <Zigara> didn't work too well on my connection
[15:10:29] <nahamu> Zigara: WAN or LAN?
[15:10:40] <Zigara> WAN, I'm sure it would be blazing fast on LAN
[15:10:58] <Zigara> I didn't do enough research to make any real comments on it though
[15:11:10] <Zigara> should try it out more sometime
[15:11:15] <KermitTheFragger> spice was supposed to be uber cool right? beating NX by a mile?
[15:11:30] <KermitTheFragger> (don't have much experience with spice)
[15:11:52] <nahamu> KermitTheFragger: I don't know. All I know is that it's *outside* the guest OS and integrated with QEMU which provides some benefit.
[15:12:07] <nahamu> If NX were better, you could always run NX inside the VM
[15:12:27] <KermitTheFragger> im currently using NX (neatx) but im running into stability issues (with all the users in one "X")
[15:12:35] <nahamu> but if you want to effectively give people "console access" even when the guest OS hasn't booted yet, you need something more like spice.
[15:12:40] <KermitTheFragger> so i thought getting them in seperate VM's might cure my headache
[15:13:00] <KermitTheFragger> nahamu: isnt that possible?
[15:13:13] <nahamu> software always cures one headache and provides a new one.
[15:13:26] <KermitTheFragger> yeah it has a habit of that...
[15:13:44] <KermitTheFragger> but there is currently no opensource NX solution that is being maintained
[15:13:55] <KermitTheFragger> and NoMachine is closing down NX with release 4
[15:14:09] <KermitTheFragger> closing up i should say
[15:14:27] <nahamu> Well, if you have a SmartOS machine, I'm happy to help you get a few test VMs up and running if the wiki instructions aren't enough.
[15:15:01] <KermitTheFragger> thx for the offer, but im not that far yet :)
[15:15:08] <KermitTheFragger> still in the exploring options phase
[15:15:29] *** arekinath has quit IRC
[15:15:38] <nahamu> Are you providing access to Windows or something else?
[15:15:45] <nahamu> and what are you using as client machines?
[15:15:50] <KermitTheFragger> centos 6
[15:16:04] <KermitTheFragger> all desktops are centos 6
[15:16:17] <KermitTheFragger> and people usually access them from home which are mixed clients
[15:16:43] <nahamu> is this centos 6 running on bare metal with neatx, or centos6 running under SmartOS?
[15:16:50] <KermitTheFragger> bare metal
[15:17:33] <nahamu> so one downside over what you're doing now is that with SmartOS you're going to pay the price of running N linux kernels rather than just one.
[15:17:53] <KermitTheFragger> yeah that is what im afraid of
[15:18:00] <nahamu> There's something called Xspice which might also be worth exploring.
[15:18:05] <KermitTheFragger> i know
[15:18:16] <KermitTheFragger> but its highly experimental
[15:18:24] <KermitTheFragger> im trying to get something more stable :)
[15:18:49] <KermitTheFragger> might always fallback to good 'ol vnc... but i doubt that will deliver an acceptable performance
[15:19:04] <Zigara> VNC is horrible for performance lol
[15:19:06] <Zigara> just so horrible
[15:19:10] <Zigara> even on LAN
[15:19:27] <KermitTheFragger> :)
[15:19:36] <Zigara> gets the job done for basic console access though
[15:19:39] <nahamu> with people connecting from home, there's always going to be a maximum number of bits you can ship to them, and certain minimum latency you'll have to pay.
[15:20:01] <jperkin> I made do with vnc from .uk->.no for quite a while, was very acceptable
[15:20:16] <jperkin> got somewhat annoying when the routing changed to go via .tx.us, but.. ;)
[15:20:49] <nahamu> I remember playing with xrdp at one point on a linux machine. definitely another thing to look at if you're shopping around for options.
[15:20:52] <Zigara> it does get the job done indeed
[15:21:12] <Zigara> as much as I don't care for windows on my servers, RDP is very nice
[15:21:21] <nahamu> then you can use rdesktop under linux and the MS rdesktop client on Mac and Windows
[15:22:03] <KermitTheFragger> sounds nice
[15:22:08] <nahamu> I don't mind Windows on my server as long as it's running as a VM under SmartOS... :-D
[15:22:23] <KermitTheFragger> ill look in to that, thanks!
[15:22:25] <nahamu> snapshot that damn thing before you make any changes.
[15:22:34] <Zigara> heh true
[15:23:19] *** denizr has joined #smartos
[15:24:45] <KermitTheFragger> though running smartos with multiple desktop images still sounds cool.. :)
[15:27:16] *** Trixboxer has joined #smartos
[15:28:04] <Trixboxer> hi, can smartos read pool created with solaris11 eval ?
[15:28:23] <jperkin> probably not, but you're welcome to try
[15:29:04] <Trixboxer> sure, if it does then worth mentioning on site ;)
[15:35:03] *** artimus has joined #smartos
[15:35:27] *** artimus is now known as Guest56088
[15:45:20] <kamilr> Hi
[15:45:33] <kamilr> Due importing image i have the following error
[15:45:35] <kamilr> Stderr from zfs receive: cannot receive: invalid stream (bad magic number)
[15:45:41] <kamilr> what does it mean?
[15:46:32] <szaydel> Your image that you are importing could be damaged.
[15:48:21] *** iyp has joined #smartos
[15:51:07] <kamilr> hm...
[15:52:28] <nahamu> where did the image come from?
[15:52:40] <kamilr> when i download image and try to install it, it goes without errors, but when i try to install it via my image server it fails BUT shasum are same
[15:52:55] <kamilr> the image cames from joyent
[15:53:15] <kamilr> but i have my own image server
[15:53:18] <kamilr> not your
[15:53:22] <nahamu> which image? which version of the platform?
[15:53:28] <kamilr> i have created in perl whith nginx
[15:54:07] <kamilr> Image - i mean one of the system image debian
[15:54:08] <kamilr> debian-6.03-1.0.0.zvol.bz2
[15:54:30] <nahamu> if you download the image file from your image server manually, the sha1sum matches correctly?
[15:54:59] <szaydel> Something is causing the stream to break, and I would guess it is the image server.
[15:55:05] <kamilr> hm....
[15:55:21] <szaydel> Is there a particular image that does not work, or have you tried more than 1?
[15:55:36] <kamilr> i will calculate the shasum on the fly and insert it into test header
[15:56:58] <iyp> Can anyone tell me the process by which settings in /usbkey/config get turned into network settings / sysinfo output?
[15:57:25] *** jamesd has joined #smartos
[15:58:11] <szaydel> I believe network stuff is all dladm and ifconfig, no other config files.
[15:58:26] <kamilr> szaydel: yes, i tried 2
[15:58:34] <kamilr> and the shasum is same
[15:58:45] <kamilr> when i computed it due request
[15:58:48] <jperkin> iyp: variables get turned into CONFIG_<var> and then used by e.g. /lib/svc/method/net-physical
[16:00:31] <iyp> jperkin: awesome
[16:01:02] <szaydel> internally net-physical relies on ifconfig as far as I recall, no?
[16:01:17] <iyp> jperkin: is there immediate reason you can think of for my admin interface not registering a vlan I specify?
[16:01:37] *** tonyarkles has joined #smartos
[16:02:31] <jperkin> iyp: none, sorry, I haven't had the need to mess with that stuff
[16:02:49] <iyp> no worries. thanks for the pointer on the setup
[16:06:19] *** denizr has quit IRC
[16:06:40] *** denizr has joined #smartos
[16:07:38] *** jamesd has quit IRC
[16:16:22] *** neophenix has joined #smartos
[16:18:41] *** CarlosC has joined #smartos
[16:23:22] *** brandon has joined #smartos
[16:23:45] *** brandon is now known as Guest20656
[16:24:11] *** porkbelt has quit IRC
[16:26:06] <EMH_Mark3> is there a way to tell how many blocks a file is using in zfs? or to see if files are using gang blocks?
[16:28:54] <iyp> jperkin: so apperantly, the admin network IP gets assigned directly to the physical interface while the external network is setup as a vnic
[16:29:04] <iyp> in the global zone on the headnode
[16:29:23] <iyp> know why this would be the case?
[16:29:34] <rmustacc> The admin network cannot have a vlan tag.
[16:29:39] <iyp> bingo
[16:29:48] <iyp> been waiting for someone to say that directly
[16:30:06] <iyp> I assume thats what The Admin network is connected in VLAN ACCESS mode only?
[16:30:06] *** bens1 has quit IRC
[16:30:13] <iyp> can only see untagged frames?
[16:32:25] <iyp> the phrase VLAN ACCESS is drawn from the SDC docs
[16:32:31] <iyp> http://wiki.joyent.com/wiki/display/sdc/SDC+Configuration+Help
[16:35:15] <iyp> rmustacc: is admin the only network with that restriction? I'm guessing it can't be an aggregation either?
[16:35:34] <rmustacc> It might be able to be an aggregation, I don't know. Ask support, they do.
[16:36:52] <iyp> rmustacc: righto. Thanks for the eye opener.
[16:39:09] *** potatosalad has joined #smartos
[16:43:40] <szaydel> iyp: http://wiki.smartos.org/display/DOC/Managing+NICs#ManagingNICs-LinkAggregationsintheGlobalZone
[16:45:30] <iyp> szaydel: Cool!
[16:45:36] <rmustacc> Please contact support for modifying the SDC config file about things like this. What's on the SmartOS wiki may not be quite right for that and we want to make sure you don't break upgrade.
[16:46:19] <iyp> Roger that. I just open a ticket through the joyent website, right?
[16:46:50] <rmustacc> Whatever process they told you to use, Sorry, not really sure.
[16:48:46] *** jwk404 has joined #smartos
[16:50:05] *** szaydel has quit IRC
[16:50:05] <iyp> rmustacc: guess that was never made clear to me. Part of the reason why I'm still bouncing around here.
[16:50:22] *** mgdm has left #smartos
[16:50:38] *** scarcry has joined #smartos
[16:50:58] <samu> Hello.
[16:51:13] <scarcry> jperkin: any insight when irssi multiarch will be fixed?
[16:51:20] <samu> I'm trying to compile some software which yields that gethostbyname is an undefined symbol... I'm a bit new to smartos, what may I do to fix this?
[16:56:56] *** nefilim has joined #smartos
[16:58:29] *** Trixboxer has quit IRC
[16:59:21] *** alucardX has quit IRC
[17:00:56] *** szaydel has joined #smartos
[17:00:57] *** tonyarkles has quit IRC
[17:03:46] <jperkin> scarcry: I already fixed it, just waiting on package rebuilds
[17:04:45] *** jaimef has joined #smartos
[17:05:31] *** bens1 has joined #smartos
[17:07:18] *** Azbruh has quit IRC
[17:08:33] <rmustacc> samu: Is it complaining during the linking phase of the compiling phase?
[17:08:45] <scarcry> jperkin: fantastic, will be looking for it later today
[17:08:49] <samu> rmustacc: linking
[17:09:00] <samu> it's ld that is complaining.
[17:09:08] *** Azbruh has joined #smartos
[17:09:19] <jperkin> samu: the manual pages will show the correct includes and libraries you need, in this case -lnsl
[17:09:30] <samu> mhm... will try, thanks
[17:10:27] <samu> jperkin: just one more thing - in `ld` manual, or `gethostbyname` manual?
[17:10:51] <samu> okay, in the symbol manual, stupid question, sorry.
[17:17:12] *** ryancnelson has joined #smartos
[17:17:13] *** scarcry_ has joined #smartos
[17:18:20] *** bluezenix has quit IRC
[17:18:20] <scarcry_> jperkin: /opt/local/etc/pkgin/repositories.conf has http://pkgsrc.smartos.org/packages/SmartOS/2012Q4-multiarch/All twice in it producing twice the output when using certain pkgin commands..are you aware or is it just me? I've created 3 zones from said uuid and all have it
[17:19:13] *** scarcry has quit IRC
[17:19:18] *** scarcry_ is now known as scarcry
[17:19:28] <rmustacc> samu: So if you look at the gethostbyname manual page you see something like this:
[17:19:52] <rmustacc> 'cc [ flag... ] file... -lnsl [ library... ]'
[17:20:00] <rmustacc> It's the -lnsl which is saying link against the nsl library.
[17:20:19] *** ryancnelson has left #smartos
[17:21:14] <jperkin> scarcry: hm, odd, there's only one in the bootstrap copy..
[17:21:55] <jperkin> but it is possible that the dataset creation script added another, I'll check it out
[17:22:05] <scarcry> jperkin: I literally used your instructions creating my zones as per your blogpost at http://www.perkin.org.uk/posts/multiarch-package-support-in-smartos.html
[17:22:29] <scarcry> only difference being my json payload files obviously
[17:24:12] <nahamu> I think I've seen that too.
[17:24:19] <nahamu> could zoneinit be doing something goofy?
[17:24:42] <jperkin> no it'll be in the creation script, it tries to detect whether the url is already set and if not adds it
[17:32:19] *** chh has quit IRC
[17:35:24] <scarcry> for those interested, I put up a wiki  on my little migration from openindiana to smartos on my main box here .. http://wiki.smartos.org/display/DOC/Migrating+from+OpenIndiana+to+SmartOS
[17:36:16] *** miine_ has joined #smartos
[17:39:11] *** dap has joined #smartos
[17:45:14] *** elijah-m_ is now known as elijah-mbp
[17:46:13] *** Guest20656 has quit IRC
[17:46:36] *** porkbelt has joined #smartos
[17:47:52] *** chrisBF has joined #smartos
[17:53:21] *** chrisBF has quit IRC
[17:58:26] *** rbrown_ has joined #smartos
[18:00:10] <rbrown_> Anyone around
[18:01:20] *** kamilr has quit IRC
[18:02:05] <Alasdairr> yes quite a lot of people are around :-)
[18:04:13] *** fortytwo_ has joined #smartos
[18:12:14] *** rbrown_ has quit IRC
[18:16:53] *** notmatt has joined #smartos
[18:24:29] *** estibi has quit IRC
[18:25:23] *** iyp has quit IRC
[18:26:54] <stetho> Hi all - I'm new to SmartOS (I started this job last week!) and I'm trying to get my head round it but I keep finding unfinished pages in the wiki - for example the Working with Packages page says "Please refer to…" Are there any other resources where I can learn about zones, installing software, security etc that might be a bit more complete?
[18:29:30] <wesolows> The Joyent docs on SDC are much more thorough, but they don't apply directly to SmartOS as they're significantly different.
[18:29:50] <wesolows> It's a wiki, so please do record your own lessons learned for the benefit of others.
[18:29:56] <jperkin> stetho: I have a bunch of posts related to packages you may find useful: http://www.perkin.org.uk/tags/smartos.html
[18:30:32] <jperkin> stetho: if you have more specific questions about zones and security we should be able to help.
[18:32:03] <stetho> wesolows: I will happily contribute to any wiki but I really am struggling to get over the first hurdles with SmartOS. It's not like the Linuxes I've used and it's been a very long time since I last saw a Solaris box. Fortunately my new employers know that but I've spent today getting not very far. :-(
[18:32:43] <stetho> jperkin: Thanks. I'll read through those.
[18:34:09] <stetho> Cool, you work for Joyent. That's the other aspect of my new job that I need to learn about!
[18:36:09] <rmustacc> stetho: What kind of stuff in particular?
[18:37:44] *** KermitTheFragger has quit IRC
[18:38:40] <stetho> rmustacc: the "particular" thing is monitoring but I'd rather learn the product than just the bits I need to know. In other words, I could come in here and say "How do I install nagios/ganglia/sensu/whatever on SmartOS", someone could give me a link and I'd achieve the immediate task but achieve nothing.
[18:38:56] *** rbrown_ has joined #smartos
[18:39:18] <stetho> *second 'achieve' should be 'learn'
[18:39:22] *** tonyarkles has joined #smartos
[18:39:44] *** Cpt-Oblivious has joined #smartos
[18:42:54] *** kamilr has joined #smartos
[18:51:03] <jesse_> jperkin, oh wow, you did a development preview dataset? ace
[18:51:15] *** rbrown_ has quit IRC
[18:52:30] <jperkin> yup, go use it ;0
[18:53:56] <jesse_> now we just need a way to switch from trunk packages to stable packages and back again...=)
[18:55:13] <scarcry> I'm currently tracking multiarch 12.4.1 .. just seems like the right repo to track for me and it needs love to make it into imgadm so ... ;)
[18:56:05] <jperkin> jesse_: that won't ever happen, unless you switch right at branch time
[18:58:09] <rmustacc> stetho: Okay, just not really sure what kind of documentation you're really looking for.
[18:59:23] <jesse_> jperkin, the problem I see is having a working install (with the trunk), and then needing some software and it requires upgrade of some library which is broken in the current trunk
[18:59:30] <jesse_> and it breaks everything
[19:00:17] <jesse_> that is, the binary packages and list of them are not persistent
[19:00:50] <jesse_> except that the list kind of is, if one doesn't run pkgin update?
[19:01:31] <jesse_> so the problem would be with older binary packages disappearing?
[19:01:43] <jesse_> (and/or versions not incrementing)
[19:03:30] <jperkin> do you have a specific example of this happening? it should in theory be ok, as you either have an outdated pkg_summary and then cannot upgrade anyway as the packages no longer exist, or you have an updated pkg_summary and then it'll tell you that the package isn't currently available.
[19:03:54] <jesse_> I mean, I install trunk now
[19:04:02] <jesse_> and it has everything I need and they work
[19:04:17] <jperkin> for quarterly releases I'll start to get rid of the rsync --delete, but with trunk that's likely to quickly eat up a lot of disk.
[19:04:18] <jesse_> and then in the future the trunk is updated and in broken state
[19:04:46] <jesse_> and I need somesoftware, but it isn't in my current pkg_summary
[19:04:55] <jesse_> so I need to do update, then install
[19:05:29] <jesse_> and somesoftware upgrades some lib package to newer version (required by it) which other software use, too
[19:05:38] <jesse_> and as the lib is broken, everything breaks
[19:05:52] <samu> Hello again
[19:06:03] <jesse_> somewhat theoretical
[19:06:21] <jperkin> that's possible, but only if someone forgets to do a revbump when upgrading the shared library, which would be bad.
[19:06:21] <samu> We're having a few zones on an OVH dedicated servers here, and we're trying to add more than one ip to a zone
[19:06:36] <samu> should I do anything more than add another 'interface' to the zone config?
[19:06:54] <samu> we've used OVH managerv3 to genrate a virtual mac for those addresses, added them to zone but they don't seem to be working.
[19:06:55] <jperkin> if e.g. libpng gets bumped then we recursively bump the PKGREVISION of everything that depends on it
[19:07:01] <jesse_> jperkin, what happens if app1 wants lib1-1.0 and app2 wants lib1-1.1 and I install app2?
[19:07:08] <jesse_> does it upgrade the lib1?
[19:08:09] *** mikl has joined #smartos
[19:08:28] <jesse_> samu, I remember seeing some discussions about that on the channel before, check the wiki for OVH references/google for the irc logs?
[19:08:31] <jperkin> yes. as to what pkgin does procedurally, I'm not sure. but, doing an 'update' before 'install' should catch it.
[19:08:53] <samu> jesse_: I've already googled the logs, but they don't say much, really ;P
[19:09:16] <jesse_> samu, the wiki, then?=)
[19:09:29] <rmustacc> samu: So they map a mac address to an ip?
[19:09:31] <jperkin> I thought we added a FAQ for that
[19:09:38] <samu> rmustacc: it looks like it does.
[19:09:44] <samu> I didn't check the vwiki, sec.
[19:09:48] <rmustacc> You should be able to add a specific mac/ip and create a new interface for the zone.
[19:09:48] <jperkin> basically, yes, you need special handling for OVH/hetzner
[19:09:53] <rmustacc> But there have been tale of woes for OVH in the past.
[19:09:57] <jesse_> jperkin, yeah, I'm just trying to figure out a situation where that libpng is not working
[19:10:08] <rmustacc> rcnelson is the man who seems to help out the most on that.
[19:10:30] *** rbrown_ has joined #smartos
[19:10:37] <samu> jperkin: you had a dedicated FAQ entry for OVH? :P
[19:10:51] <jperkin> I thought someone wrote one the last time this was discussed
[19:11:01] <rmustacc> samu: It hsa been that bad, I'm sorry to say.
[19:11:04] <rmustacc> *has
[19:11:24] <samu> I can't google anything about hetzner/ovh on smartos wiki, so I guess it was added using different text
[19:12:29] *** iyp has joined #smartos
[19:12:45] <samu> 2013-02-14.log:17:02< colstrom> I'm going to summarize everything and roll it into a "How to get your KVM Guest on SmartOS connected to OVH's network" guide.
[19:12:53] <samu> I guess you're talking about this guy ;P
[19:13:05] <jperkin> yeh, sounds about right
[19:13:52] <samu> I can't find any solution in these logs, though
[19:14:06] *** kamilr has quit IRC
[19:15:10] <rbrown_> does smartos use qemu or kvm ?
[19:16:12] <jesse_> "yes" is probably the right answer to that one?
[19:16:40] <jperkin> yeh we use both country AND western over here.
[19:17:47] <rmustacc> rbrown_: Yes, we use both.
[19:19:16] <nahamu> is there a simple way to get SmartOS to pass "localtime" to a KVM VM?
[19:19:42] <nahamu> e.g. I run a windows VM and for some reason can't use the registry hack
[19:19:52] <nahamu> (to tell windows that the hw clock is UTC)
[19:20:10] <nahamu> or do I need to use a funky startvm.zone file that sets the TZ variable?
[19:20:50] <nahamu> also, would that be a sane feature to add to vmadm? "local_timezone":"US/Eastern" triggering setting the TZ and telling QEMU to use that time rather than UTC?
[19:24:50] <xenol> Hi, I have multiple ISOs stored under /opt/isos and I would like to make them available via lofs into KVM zone root directory. However, the VM fails to start. Any ideas how could I have multiple ISOs available, but stored only once on the disk?
[19:25:04] <rmustacc> nahamu: Not that I know of.
[19:25:10] <rmustacc> xenol: How is it failing?
[19:25:33] *** iyp has quit IRC
[19:25:37] *** ryancnelson has joined #smartos
[19:26:06] <nahamu> xenol: what's the snippet of json you're using to add them in via lofs?
[19:28:56] <xenol> I have added it via zonecfg :]
[19:29:11] <samu> shit, I'm never going to do those addresses ;P
[19:29:15] <xenol> (I guess that's not the best way to do it)
[19:31:40] <nahamu> xenol: doing it that way isn't supported. if you look in the vmadm man page, there's a "filesystems" section you can use to get a filesystem passed through via lofs into the zone.
[19:32:00] <xenol> nahamu: I will check that
[19:32:46] *** enmand has quit IRC
[19:33:05] *** ismell has quit IRC
[19:34:12] <jesse_> nahamu, there's no update for filesystems
[19:34:31] *** tonyarkles has quit IRC
[19:34:47] <xenol> nahamu: I am not sure if it works for KVM. According to man page, every filesystem.* option is just for the OS vms
[19:34:52] *** rbrown_ has quit IRC
[19:34:57] <nahamu> xenol: oh, hrm...
[19:35:00] <nahamu> that's a shame.
[19:35:00] *** bluezenix has joined #smartos
[19:35:37] <nahamu> it would be a useful feature for storing a bunch of ISOs in the GZ and making it relatively easy to pass them through to a VM...
[19:35:40] <nahamu> alas
[19:37:07] *** denizr has quit IRC
[19:42:13] <xenol> nahamu: any way how to solve this?
[19:44:12] <nahamu> xenol: I don't know.
[19:44:30] <nahamu> there might be a sneaky way to do it.
[19:45:46] <nahamu> if you create a zfs filesystem with a UUID /zones/<UUID>, create a "root" directory under that, put the isos somewhere under the "root" directory, and then use that UUID as the "zone_dataset_uuid" in the json for the VMs, it might do the right thing
[19:46:14] <nahamu> it would snapshot and clone the filesystem, so it wouldn't pick up new ISOS as you add them later, but it *would* see all the isos that were there when you created the VM
[19:46:45] <ryancnelson> there had damn-well-better *already* be a /zones/<uuid>/root/ directory there, or you did something 100% crazy
[19:47:39] <nahamu> I think that's been renamed to "image_uuid"
[19:48:51] <nahamu> ryancnelson: not for an existing VM.
[19:48:59] <nahamu> that was unclear
[19:49:38] <nahamu> ryancnelson: an existing VM should certainly have a /zones/<uuid>/root/ directory
[19:50:05] <nahamu> let <ISOS-UUID> stand for a special uuid chosen as the place to hold isos
[19:51:01] <nahamu> I'm suggesting that if one did "zfs create zones/<ISOS-UUID>; mkdir /zones/<ISOS-UUID>/root; cp /zones/<old_isos_dir>/*.iso /zones/<ISOS-UUID>/root"
[19:51:47] <nahamu> then you could pass "image_uuid": "<ISOS-UUID>" in the json for a VM and be able to "vmadm boot <new-vm-uuid> cdrom=/foo.iso,ide"
[19:51:56] <nahamu> (I think. this is all speculation...)
[19:51:58] <samu> huh. with allow_ip_spoofing=1 all ips work fine on the zone..
[19:52:01] <samu> dirty fix for now.:P
[19:52:35] <nahamu> I'm piggybacking on the feature I leverage fo slipping in my spice-enabled QEMU
[19:52:39] <nahamu> *for
[19:53:26] <xenol> I have /zones/<UUID>/root created automatically. My issues is that I need to have large amount of isos available to every virtual machine
[19:53:53] <nahamu> I'm not sure how much more clearly I can describe this.
[19:54:02] <xenol> it will consume much more space than via lofs mount
[19:54:17] *** ismell has joined #smartos
[19:54:21] <nahamu> not the way I described. It will use zfs snapshots and clones.
[19:54:42] <nahamu> if you need to keep updating each zone with new isos, though, then my way won't work.
[19:56:20] <xenol> Thanks, that might work to some degree
[19:58:56] <nahamu> It should certainly merit trying with 1 or two isos to see if it works at all.
[19:59:36] *** tonyarkles has joined #smartos
[20:06:20] <samu> how can I chang a zones hostname?
[20:06:23] <samu> change*
[20:06:44] <wesolows> its nodename, or the hostname of an interface?
[20:06:50] <wesolows> the nodename should be in /etc/nodename; it's just a file
[20:07:00] <wesolows> to change hostnames, you need to modify your DNS server.
[20:07:03] <samu> # hostname
[20:07:04] <samu> 70-54-d2-19-6c-f1
[20:07:06] <samu> i'm talking about this ;P
[20:07:08] <wesolows> that's the nodename
[20:07:18] <wesolows> a machine, virtual or otherwise, has one nodename
[20:07:39] <wesolows> it may have anywhere from 0 to a very large number of hostnames, which are just name-to-address mappings.
[20:08:11] <samu> okay, now, can I change it withoug restarting?
[20:08:11] <wesolows> the hostname(1) command dates back to an era in which each system had exactly one :-/
[20:08:12] <scarcry> wesolows: if the mac address of the admin nic in the GZ is a nodename, why does /etc/nodename contain unknown?
[20:08:24] <samu> okay, worked.
[20:08:25] <samu> thanks.
[20:08:25] <samu> ;P
[20:08:37] <wesolows> samu: svcadm restart identity:node, I think
[20:09:13] <wesolows> scarcry: I'm not sure.
[20:09:19] <wesolows> the GZ is special though
[20:09:25] <scarcry> sure is
[20:09:39] *** tonyarkles has quit IRC
[20:09:44] <wesolows> there are many ways to set the nodename on a running system
[20:10:32] <wesolows> they all end up calling sethostname(3c) though
[20:14:27] *** denizr has joined #smartos
[20:16:08] *** tonyarkles has joined #smartos
[20:17:35] <samu> now, just last one question and I'll leave you alone ;P
[20:18:20] <samu> I see that /usr is a read-only filesystem, but is it possible to add a file there somehow? I'd like to make a symlink from /usr/home to /home, because we've migrated this server from freebsd and that would help many users a lot with their configs
[20:18:24] *** bens1 has quit IRC
[20:18:40] <wesolows> the only way would be to rebuild the platform image
[20:18:58] <wesolows> take a look at what smartos-live does; you'll need to hook into that, or write your own
[20:20:54] <wesolows> for zones, note that you could do what we do with the fake-subset: copy /usr into /opt/whatever, then mount that over /usr.  Extremely gross, but probably possible.
[20:21:01] <samu> mhm. okay, then I guess I won't touch that ;P
[20:21:44] <wesolows> you may want to look into sngl; that's a writable-/usr option for zones.
[20:21:49] <wesolows> still experimental
[20:22:24] <samu> thanks, but I guess I'll just write about the path difference in the post-migration mailing. Thanks anyway ;0
[20:22:27] <samu> ;)
[20:26:19] *** tonyarkles has quit IRC
[20:28:11] *** tonyarkles has joined #smartos
[20:29:46] *** iyp has joined #smartos
[20:30:12] *** bluezenix has quit IRC
[20:31:15] *** ipalreadytaken has quit IRC
[20:31:24] *** ipalreadytaken has joined #smartos
[20:33:40] *** bluezenix has joined #smartos
[20:34:13] *** notmatt has quit IRC
[20:34:29] *** bluezenix has quit IRC
[20:35:45] *** ipalreadytaken has quit IRC
[20:37:13] *** potatosalad has quit IRC
[20:38:03] <jesse_> is there a way to tell vmadm create 'yes I want to create a zone with same IP as a stopped zone'?
[20:39:28] *** potatosalad has joined #smartos
[20:40:54] *** iyp has quit IRC
[20:41:25] <nahamu> jesse_: I don't think so. bad things would happen if you then booted the stopped one
[20:41:53] <jesse_> well, put a check in start, then=)
[20:41:55] <nahamu> why do you want to keep multiple zones all configured with the same IP?
[20:42:02] <jesse_> dev/test zones
[20:42:07] <jesse_> only one running at a time
[20:42:22] <jesse_> archive of old setups/environments
[20:42:40] <jesse_> disk is cheap, ips are not=)
[20:42:43] <nahamu> start just tells zoneadm to start the zone. you'd have to add a lot of extra brains to a step that is currently dead simple.
[20:43:16] *** iyp has joined #smartos
[20:44:14] <nahamu> if you want your own version of the platform that gives you full power to shoot yourself in the foot, this is approximately the patch you would want: https://github.com/nshalman/smartos-live/commit/14dea8a30dd6b1fc043b6ca67176152875292fe8
[20:44:15] <jesse_> move the check logic from create to start
[20:44:24] <nahamu> jesse_: there is no logic in start
[20:44:26] <nahamu> that's my point
[20:44:46] <nahamu> you'd have to *create* logic in start
[20:44:54] *** iyp has quit IRC
[20:45:52] <jesse_> now, maintaining a separate smartos for the dev box is more of a hassle than just vmadm update delete/add interfaces...
[20:47:15] <jesse_> I don't see how it would be hard to add check to 'vmadm start' that is already in 'vmadm create'
[20:47:28] <jesse_> without looking at the code
[20:47:42] <nahamu> smartos-live/src/vm/node_modules/VM.js
[20:47:44] <jesse_> and I'm assuming sane quality of code here=)
[20:48:05] <nahamu> https://github.com/joyent/smartos-live/blob/master/src/vm/node_modules/VM.js
[20:48:16] <nahamu> If there's a sane way to do it, open a pull request. :)
[20:48:17] <jesse_> (actually, I think I've gone through that code when I was debugging something else hald a year ago)
[20:48:22] <jesse_> half, too
[20:49:24] <nahamu> https://github.com/joyent/smartos-live/blob/master/src/vm/node_modules/VM.js#L7992
[20:50:23] <nahamu> one might argue, though, that if a zone exists, it should always be bootable.
[20:50:50] <nahamu> whereas you argue that a zone that exists should be bootable only if it doesn't use an IP that's already in use.
[20:51:19] <jesse_> one might argue, that if a zone is started, it shouldn't set autoboot=true, either
[20:51:55] <nahamu> I think their reasoning is probably that if it's running, it should come back up on a reboot.
[20:52:04] <jesse_> you assume the zones are long lived and persistent
[20:52:04] <nahamu> I think stopping might alter autoboot as well.
[20:52:18] <nahamu> jesse_: but that's exactly how Joyent uses SmartOS.
[20:52:45] <jesse_> yes, but it doesn't mean smartos has to do it that way, too=)
[20:53:17] <jesse_> anyway, hacking the system is not a solution this time, so hell-of-vmadm-updates it is
[20:57:46] <nahamu> jesse_: there is one hack you could consider, though I don't recommend it...
[20:57:56] <nahamu> totally unsupported...
[20:57:58] <jesse_> ...edit the zone xml files?=)
[20:58:08] <nahamu> I was going to say move them out of the way
[20:58:17] <nahamu> and then put them back when you want the zone back.
[20:58:42] <nahamu> you might have to update the index file too... I'm not sure.
[20:58:57] <jesse_> I wonder if vmadm update does the same check as create...=)
[21:07:40] <jesse_> (yes, it does)
[21:10:54] *** bluezenix has joined #smartos
[21:13:30] *** Teknix has quit IRC
[21:15:53] *** Teknix has joined #smartos
[21:16:21] *** ipalreadytaken has joined #smartos
[21:20:33] *** tonyarkles has quit IRC
[21:28:05] <jesse_> jperkin, still no working nsd in trunk? The newest one compiled with ./configure && make just fine...
[21:29:03] <jperkin> hm, thought I fixed that..
[21:29:10] <jesse_> (newest meaning nsd-3.1.14)
[21:29:41] <jperkin> => Returning to build of nsd-3.2.15
[21:30:06] <jperkin> why the version mis-match?
[21:30:41] <jesse_> because I mistyped, 3.2.14 is the one I compiled in january
[21:31:03] <jperkin> ok :)
[21:32:53] <jesse_> trunk has the libxcb fix, too?
[21:33:21] <jesse_> hmmm, still no cups
[21:33:28] <jperkin> ok, I just fixed nsd upstream, it'll make its way into trunk in a couple of days
[21:33:31] <jesse_> annoying dependency in openjdk, it only needs the header=)
[21:33:56] <jperkin> yes trunk should have the libxcb stuff, I upstreamed that
[21:35:24] <jesse_> I wonder what else might still be missing
[21:36:28] *** pringlescan has joined #smartos
[21:36:31] <jperkin> quite a lot as it's based on multiarch, http://pkgsrc.smartos.org/reports/trunk/20130301.1729/meta/report.html is the most recent report
[21:36:42] *** wramthun has quit IRC
[21:37:05] <jperkin> but should be no less than our current repos
[21:37:57] <jesse_> thinking in lines of compile-java and what-I-run-personally
[21:38:03] *** wramthun has joined #smartos
[21:38:04] <pringlescan> Hello all, I'm trying to setup Smart OS on a baremetal server and it reboots when I try to start it up off a USB drive after it says "Syncing file systems…" it does have a RAID card installed, it's a Dell R710. I'm not sure how to go about trouble shooting.
[21:39:21] <jperkin> jesse_: give it a spin, let me know if there are particular packages you need and I can put them at the top of the queue
[21:39:35] <jesse_> already testing
[21:39:38] *** scarcry has quit IRC
[21:40:00] <jesse_> so far nsd was the only thing missing
[21:40:39] <jesse_> time to test the openjdk build
[21:40:59] <jesse_> painful to set up
[21:44:36] *** scarcry has joined #smartos
[21:44:58] <jesse_> hmmm, no sun-jre6 or sun-jdk6 in the repository
[21:45:46] <jperkin> it's trunk so it tries to use the native pkgsrc java
[21:46:07] <jperkin> which I happen to have added support to a few days ago, so the next build should pick it up
[21:46:09] <jesse_> which is kind of fatal, as those are required to compile the native java0)
[21:46:32] <jesse_> (or another jdk binary set)
[21:46:34] *** ipalread_ has joined #smartos
[21:46:44] <jesse_> did you use the set I compiled as the binary seed?
[21:47:20] *** notmatt has joined #smartos
[21:47:33] <pringlescan> How can I find out if my RAID card is supported by the latest USB image? I'm having trouble finding this info for SmartOS in particular?
[21:47:56] <jesse_> pringlescan, illumos hardware compatibility list
[21:48:17] <pringlescan> http://wiki.smartos.org/pages/viewpage.action?pageId=755673 this lists my exact server, but doesn't specify the RAID card in use
[21:49:29] <jperkin> jesse_: no, this is for using the sun^Woracle jre/jdk, previously we had a pkgsrc-joyent package for that as the pkgsrc one only supported the linux version (which is what *BSD uses via emulation)
[21:49:32] <ryancnelson> really?  googling "smartos hardware requirements" doesn't show you?
[21:49:37] *** ipalreadytaken has quit IRC
[21:49:47] <pringlescan> ryancnelson, nope smart os raid compatibility didn't give me anything
[21:50:00] <ryancnelson> 'cause it's spelled "smarts"
[21:50:02] <ryancnelson> smartOS
[21:50:04] <pringlescan> I google about 5 or 6 different ways and OpenIndiana has better SEO
[21:50:08] <ryancnelson> stupid autocorrect
[21:50:34] <pringlescan> I didn't catch Google autocorrecting it :-(
[21:51:09] *** notmatt has quit IRC
[21:51:10] <jesse_> jperkin, ah, true, only the joyent set had jdk 6 before, too
[21:51:21] <pringlescan> So, my hardware is supported, I followed the official instructions for making a USB boot drive, and when I try to boot it restarts my machine when it prints "syncing filesystems" to the console. Any ideas?
[21:51:41] <jesse_> jperkin, another missing: apache-ant (probably because it depends on the jre)
[21:51:56] <jperkin> yeh, that will pick up a bunch of dependencies
[21:52:04] <jperkin> which is why I made that change, should get ~100 more packages.
[21:52:10] <pringlescan> I just saw it flash again it does say …. Done after syncing file systems; then it reboots.
[21:52:18] * jperkin is gunning for 10,000 packages this year ;)
[21:52:47] <pringlescan> I think it's related to this: https://www.illumos.org/issues/3287
[21:53:15] <ryancnelson> why do you suspect your hardware is supported?  what perc card do you actually have?
[21:53:51] *** bluezenix has quit IRC
[21:54:00] <jesse_> jperkin, =)
[21:55:50] <pringlescan> That ticket refers to a storage controllers page which isn't in the illumos.org wiki, are they referring to the one on open indiana? I'm finding this to be somewhat counter-intuitive and I've used Confluence extensively.
[21:57:10] *** notmatt has joined #smartos
[21:57:26] *** wramthun has quit IRC
[21:57:44] <nahamu> You could look in the https://www.illumos.org/hcl/ and perhaps boot the machine in "noinstall" mode to verify the PCI ID.
[21:57:52] *** stetho has quit IRC
[21:58:45] <pringlescan> noinstall dies too
[21:59:02] *** wramthun has joined #smartos
[21:59:03] <nahamu> pull the raid card and see if that fixes the problem?
[21:59:32] <jesse_> jperkin, idea when the next build might be done?
[21:59:35] <nahamu> if it boots happily without the raid card and reboots with it, you'll probably have found the culprit.
[22:00:00] <pringlescan> it's not an option not to boot with the raid card though, so I'm not sure if I want to take that troubleshooting step
[22:00:51] <nahamu> Dell PERC cards are a bit notorious from what I understand.
[22:00:51] <pringlescan> this card is from 2007… so I'm guessing my new box with 2012 hardware won't be supported either… I was really excited about SmartOS but it seems to be dead in the water for my purposes and the documentation is spread out between illumos, smartos and openindiana correct?
[22:01:37] <pringlescan> nahamu, one is a Dell PERC card, works fine in ESXi and Ubuntu, not that they're comparable to Smart OS… the card coming is also LSI-based
[22:01:42] *** porkbelt has quit IRC
[22:02:08] <pringlescan> Illumos and OpenIndiana say that  PERC 6/i cards are supported though
[22:02:10] <rmustacc> pringlescan: Boot -k.
[22:02:17] <rmustacc> Add that to your prompt.
[22:02:24] <rmustacc> That should cause it to stop at the panic.
[22:02:26] <pringlescan> rmustacc, no install?
[22:02:44] <jperkin> jesse_: current build started 6 hours and should have the java changes in, that will complete probably in a day or so (I made a change to gcc47-libs so pretty much every package will get a rebuild)
[22:02:45] <rmustacc> No, boot the same way that you are, but add -k to the boot options.
[22:03:07] <pringlescan> ok, I've never booted this is my first try
[22:03:08] <rmustacc> That should cause you to halt into the debugger instead of rebooting when we panic.
[22:03:26] <pringlescan> I'm assuming the install will let me create my storage pool and then I continue to boot off of usb
[22:03:28] <rmustacc> Then you can share the message and we can make forward progress.
[22:03:32] <rmustacc> It will.
[22:03:39] <rmustacc> But as you said, you're not getting far enough to install.
[22:03:52] <jesse_> jperkin, ok, I'll check it Thursday/Friday
[22:04:01] <jesse_> should be done by then
[22:04:05] <pringlescan> I wish this machine didn't take 3-5 minutes to reboot
[22:04:21] <jperkin> jesse_: yep, a full build takes no more than 2 days
[22:05:21] <jesse_> openjdk will add another 30-40 minutes to that. If you have 12c/24t xeon to compile it with=)
[22:05:52] <jperkin> it's parallel across 6 machines
[22:06:02] <jperkin> MAKE_JOBS=8 on each
[22:06:05] <jesse_> actually, I think only hotspot uses parallel compiling
[22:06:11] <jesse_> rest is single-threaded
[22:06:13] <pringlescan> rmustacc, -k worked, now what should I do, it says "no dump device" and I'm at a [0]> prompt
[22:06:55] <rmustacc> Did it print a lot of messages above that?
[22:07:00] <rmustacc> eg. a stack trace, etc.
[22:07:54] <pringlescan> Yes, it does
[22:07:56] <jperkin> scarcry: 2012Q4-multiarch has irssi now.
[22:08:27] <scarcry> I was just looking at http://pkgsrc.smartos.org/reports/2012Q4-multiarch/20130304.1727/meta/report.html
[22:08:56] <pringlescan> after some ACPI, we get to post_startup and it dumps at genunix:main and unix:_locore_start
[22:09:09] <pringlescan> I can take a photo of the screen
[22:10:22] <rmustacc> Can you please?
[22:11:28] *** scarcry_ has joined #smartos
[22:11:57] <pringlescan> rmustacc, thanks: http://imgur.com/G7bNMkY
[22:12:46] <scarcry> jperkin: thanks a lot, seems like I'm back in business with irssi perl support :)
[22:13:10] *** ipalreadytaken has joined #smartos
[22:13:19] *** scarcry has left #smartos
[22:14:02] <jperkin> sweet. now I need to fix irssi-{icb,xmpp} ;)
[22:14:56] *** scarcry_ is now known as scarcry
[22:15:51] *** ipalread_ has quit IRC
[22:17:48] <pringlescan> How hard it is to manually compile packages on solaris based machines versus Linux and BSD?
[22:18:13] <jperkin> pringlescan: usually more work, that's why we have pkgsrc
[22:18:45] <pringlescan> I don't run anything too exotic and if I do it's usually something small off GitHub with few dependencies.
[22:19:22] <wesolows> it varies a lot from one thing to the next, depending on how arrogant the author is
[22:19:31] <wesolows> you may substitute "ignorant" if you prefer.
[22:19:36] <pringlescan> If I get this installed is there a good book for people with a linux background transitioning to solaris?
[22:19:41] <rmustacc> pringlescan: That one is going to be hard.
[22:19:45] <wesolows> The stuff from the FSF, for example, tends to be easy to build anywhere.
[22:19:53] <pringlescan> rmustacc, my kernel panic?
[22:20:21] <pringlescan> I disabled C-states in the bios, I'm going to disable turbo mode too and maybe virtualization and see if that does anything… just for lack of having a better idea of what to do
[22:20:32] <pringlescan> holy shit
[22:20:41] <pringlescan> … good guess on my part, it booted haha
[22:20:54] <rmustacc> Oh, you disabled C-states and that worked?
[22:21:05] <pringlescan> yes, I don't remember if I disabled turbo as well, I believe not
[22:21:07] <rmustacc> You said this was a Dell machine?
[22:21:26] <rmustacc> What generation is that?
[22:22:41] <pringlescan> It's an R710, I'm not sure, also, I can't update the BIOS because Windows won't install on it and the Cent OS iso I found online wouldn't work either.
[22:23:00] <rmustacc> Well, if it works with C-states disabled, that's good.
[22:23:04] <pringlescan> I think it's just an update for iDRAC, but I'm disabling that because it's full of security holes apparently.
[22:23:22] <wesolows> the thing is that C6 was completely broken in Nehalem and possibly also Westmere, so it wouldn't be surprising if the ACPI tables for it weren't tested well
[22:24:01] <wesolows> I believe C6 and C7 are now safe if you have Sandy Bridge/Patsburg.
[22:24:05] <pringlescan> it appears to be working I did get a cp: error about sources.list but it's rebooting so I couldn't see it
[22:24:16] <wesolows> ... but not strongly enough to enable them on our systems. :-)
[22:24:44] <pringlescan> do I need c-states? it'll be in a data center, I'm not paying for power
[22:24:49] <ryancnelson> you do not
[22:24:59] <wesolows> well, yeah, I mean, you need C0.
[22:25:02] <pringlescan> now when I reboot the second time, do I pick no install since I just setup storage
[22:25:07] <wesolows> you don't need anything below C1E.
[22:25:31] <wesolows> a system without C0 is going to be of little utility.
[22:25:46] <wesolows> fortunately these "BIOS options" are typically around C3 or C6/C7 support only.
[22:25:53] <wesolows> which can be safely disabled
[22:26:05] <pringlescan> The issue is supposedly fixed in 6.3.0 of the BIOS
[22:26:10] <wesolows> that's nice.
[22:26:39] <wesolows> R710 is EOL, so I would suggest finding a BIOS + settings that work and never touching any of them again.
[22:27:06] <pringlescan> I just don't understand what the point of iDRAC is if the auto-update doesn't work..
[22:27:30] <wesolows> the point of idrac is to make money for dell.
[22:27:54] <pringlescan> I didn't order from them for my new server, I'm a mile from the datacenter, I don't need remote access.
[22:28:11] <pringlescan> ASA computer had reasonable prices
[22:28:20] <wesolows> well, most of the work we're doing these days assumes that you have serial console access to the system.
[22:28:40] <wesolows> because the reality is that remote management (specifically, console) is a basic fundamental feature of DC-quality hardware.
[22:28:53] *** Guest56088 has quit IRC
[22:28:58] <pringlescan> can I get serial output from a usb to serial converter and boot smartos ttya?
[22:29:10] <wesolows> I doubt it.
[22:29:20] <wesolows> Certainly GRUB will not work over such an interfacce.
[22:29:54] <wesolows> Note that VGA/keyboard is still supported, it's just not an interface that is recommended or used by most people.
[22:30:13] <pringlescan> I'm about to just get a raspberry pi and use that as PXE server
[22:30:21] <pringlescan> so how do you do it? (get serial access)
[22:30:37] *** estibi has joined #smartos
[22:30:53] <pringlescan> … ok so unless the setup just overwrote my flash drive (it only listed one available drive)… it no longer boots
[22:31:03] <pringlescan> I assumed it found my disk array, I guess that was a flawed assumption
[22:31:03] <wesolows> There are two approaches.  The first, and by far the most common, is to use the serial redirection in the server's BMC, set up to stop redirection on the bootloader start.
[22:31:22] *** bens1 has joined #smartos
[22:31:28] <wesolows> The second is to use a terminal concentrator, which won't allow you access to the BIOS/firmware stuff, but will work fine with GRUB and the OS.
[22:32:15] <pringlescan> I only have two servers and again, a mile from the datacenter, so that seems like overkill, but it would be helpful for troubleshooting issues like this.
[22:32:32] <pringlescan> so i guess my raid still wasn't detected, and smartos made a zfs pool on my flash drive
[22:32:54] <wesolows> what does diskinfo show?
[22:34:39] <pringlescan> I'll have to reformat the disk
[22:34:43] <pringlescan> it rebooted and now it won't boot
[22:34:47] <pringlescan> *re-image the disk
[22:34:57] <pringlescan> that'll take about 30 minutes
[22:35:24] <pringlescan> I think I can set up a PXE server faster
[22:38:17] *** ipalreadytaken has quit IRC
[22:39:31] *** artimus has joined #smartos
[22:39:45] *** ipalreadytaken has joined #smartos
[22:39:53] *** artimus is now known as Guest92874
[22:43:05] <pringlescan> wesolows, the flash drive has a /boot and a /platform folder on it still, I wouldn't have a /boot platform if it wiped my disk, so it just stopped booting for some reason
[22:43:31] <pringlescan> also, it formatted my RAID definitely because the O/S on there no longer boots… any ideas for troubleshooting this :-\
[22:43:45] *** Guest92874 has quit IRC
[22:43:59] *** wolfeidau has quit IRC
[22:46:18] <pringlescan> I set the partition active again just to check, I guess I'll reflash the USB, not sure why I'd need to though
[22:46:58] <wesolows> sorry, I don't understand the problem.
[22:47:25] <pringlescan> I ran the setup, it rebooted, and now it won't boot off the USB.
[22:47:30] <ryancnelson> ah
[22:47:42] <ryancnelson> i bet you're set to try to boot from disk first
[22:47:53] <pringlescan> I'm manually setting it to USB
[22:48:00] <wesolows> oh, the firmware might be seeing the GPT on the disk and flipping out, yeah
[22:48:04] <ryancnelson> on a 710?
[22:48:09] <pringlescan> yes, on a 710
[22:48:31] <ryancnelson> ... those boot from "disk", ... and disk is defined as "this disk, or this stick"... if you pull the stick, it resets
[22:48:41] <ryancnelson> ... or are you hitting f11 to pick manually?
[22:48:46] <pringlescan> F11 to pick manually
[22:49:22] <pringlescan> It worked when I boot up the setup
[22:51:02] <pringlescan> o_O
[22:51:22] <pringlescan> I just tried for the 10th time (I've tried different ports), cold-reboot, etc, and all of a sudden it worked on the 3rd try on this boot
[22:51:48] <pringlescan> I really think I need to get this BIOS updated, but I tried for a whole day, it has a ton of issues booting from USB and the optical drive won't work at all in the BIOS but works from Linux
[22:52:31] <pringlescan> yay! I have a login prompt =) thanks guys… it appears to be very buggy hardware causing my problems
[22:54:01] <nahamu> I hope that hardware is just for testing...
[22:54:10] <nahamu> I wouldn't trust it in production...
[22:55:57] *** bens1 has quit IRC
[22:56:14] <pringlescan> it's for a backup server but it's in the DC as a backup, so it needs more testing I guess
[22:56:36] <pringlescan> or I should just run ESXI on it because I won't hit the ram limit I will on my main box but I wanted them to be identical configurations
[22:58:15] *** szaydel has quit IRC
[23:01:38] *** ryancnelson1 has joined #smartos
[23:01:58] *** bluezenix has joined #smartos
[23:02:41] *** wolfeidau has joined #smartos
[23:03:10] *** ryancnelson has quit IRC
[23:04:02] *** JT-EC has quit IRC
[23:04:10] *** JT-EC has joined #smartos
[23:06:44] *** szaydel has joined #smartos
[23:11:34] *** tonyarkles has joined #smartos
[23:12:00] *** bluezenix has quit IRC
[23:13:26] *** denizr has quit IRC
[23:15:49] *** CarlosC has quit IRC
[23:18:17] *** ChrisPartridge has joined #smartos
[23:19:48] *** ipalreadytaken has quit IRC
[23:25:12] *** ipalreadytaken has joined #smartos
[23:36:27] *** ipalreadytaken has quit IRC
[23:37:29] *** bluezenix has joined #smartos
[23:37:55] *** ipalreadytaken has joined #smartos
[23:48:21] *** ryancnelson1 has quit IRC
[23:55:17] *** porkbelt has joined #smartos
[23:56:18] *** ryancnelson has joined #smartos
top

   March 4, 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 | 31 | >