[00:00:03] *** Orpheus has quit IRC
[00:05:06] <Saskaloon> IPMI, like iLO, ILOM, DRAC, or iKVM?
[00:05:46] <Saskaloon> ... and VMX?
[00:06:16] <rmustacc> IPMI like those.
[00:06:38] <rmustacc> VMX is the name of the Intel instruction set extension.
[00:06:45] <rmustacc> It'll become a bit of a friend.
[00:07:09] <rmustacc> As will Intel Architecture manual 3B
[00:14:16] <Saskaloon> I had bought a Xeon W3680 for a processor; but, I paired it with a P6T6 - no IPMI.
[00:14:44] <rmustacc> It's not the end of the world.
[00:16:49] <Saskaloon> Whoa, 16.5MB PDF from Intel there...
[00:17:59] *** luuse has quit IRC
[00:18:11] *** brendang has quit IRC
[00:19:04] <rmustacc> Yes, ypu're looking at 1000s of pages of docs.
[00:19:40] *** brendang has joined #smartos
[00:20:26] *** knweiss has joined #smartos
[00:21:29] <Saskaloon> Looks like I got an omnibus edition - 4175 pages.
[00:21:39] <Saskaloon> Volumes 1 through 3c.
[00:24:32] <Saskaloon> SmartOS supports non-global illumos zones (ie. non-kvm), as well, right?
[00:24:44] <e^ipi> yes, of course
[00:25:31] <Saskaloon> Maybe I just need to rtfm; but, how do the ng zones get updated, when you boot from a newer USB drive?
[00:26:11] * Saskaloon goes to the wiki...
[00:26:35] <rmustacc> Well, the NGZ is comprised of two parts: The sparse root and the pkgsrc packages.
[00:26:57] <Saskaloon> Oh, no whole-root ngz's, then.
[00:27:29] <Saskaloon> Well, that would certainly do it.  :)
[00:29:51] <rmustacc> Nope, none.
[00:31:05] <rmustacc> So the sparse zones always get updated on the reboot.
[00:31:24] <rmustacc> The idea is that the only things that are changing in the platform are things which are guaranteed to be compatible from a linking/library perspective.
[00:31:30] <rmustacc> Otherwise you're using the version from pkgsrc.
[00:31:43] <rmustacc> And that has its own orthogonal update process.
[00:31:49] <Saskaloon> Or downgraded, if you revert to a previous version?
[00:33:09] <rmustacc> If you haven't used any newer features, yes.
[00:34:02] <rmustacc> Libraries have versioned symbols.
[00:34:31] <rmustacc> So if you haven't used something that's in the newer version, but not in the older one, then you're not going to work.
[00:34:54] <Saskaloon> Got it.
[00:34:55] <rmustacc> That's true of all the core libraries that come from illumos.
[00:35:47] <Saskaloon> So, "fix and move-forward" would be the ideal development path.
[00:36:13] <rmustacc> Hmm?
[00:36:19] <killfill> there should be a way to mark the external network as the default one, right?
[00:36:26] *** ryancnelson has joined #smartos
[00:38:14] <Saskaloon> Just thinking, downgrading should be avoided, upgrading is mainly fine though.
[00:39:09] <killfill> $ sdc-createmachine -> At least one network must be assigned to one package
[00:39:14] <rmustacc> Yeah, something like that. I would say it's okay to roll back if something pops up right away after an upgrade, but other than that.
[00:39:16] <ryancnelson> ah
[00:39:23] <ryancnelson> killfill: you're on sdc?
[00:39:28] <killfill> ryancnelson: yup
[00:39:32] <rmustacc> killfill: You need to manually go into the packages and add the networks you want on them.
[00:39:37] <ryancnelson> you can associate a network with a package
[00:39:43] <ryancnelson> … but out of the box, they don't have one
[00:39:57] <ryancnelson> edit your package, or create a new one
[00:40:17] <ryancnelson> … and then when you pick that package, it implies the network(s) that go with it
[00:40:27] <killfill> Aah greate
[00:43:08] <killfill> how do i wipe out destroyed machines? "No available servers with this dataset"
[00:43:31] <killfill> is it a space thing?...
[00:44:16] <ryancnelson> destroyed servers still show in the portal, as "destroyed", but they are really gone.
[00:44:30] <ryancnelson> that's just a record of what has been there before.
[00:44:35] <killfill> ah
[00:44:53] <ryancnelson> it's designed to be a public cloud… so you need to know what a customer *had* as well as *has now*
[00:45:21] <killfill> there has to be something else then.. well i just have a head node.. in the ops panel i select headenode as the compute node.. :P
[00:45:23] <ryancnelson> … for billing, and to know things like "who had this IP address last week, when it was doing evil things"
[00:45:29] <killfill> sure
[00:46:03] <ryancnelson> "no servers with this dataset" means it can't find a place to put that vm that doesn't break some other rule
[00:46:12] <ryancnelson> how many compute nodes do you have?
[00:46:23] <killfill> none.. just the head one
[00:46:24] <killfill> :P
[00:46:35] <ryancnelson> is it marked as "reserved"?
[00:46:41] <ryancnelson> (it usually is"
[00:46:46] <ryancnelson> )
[00:47:09] <ryancnelson> … so you can provision there manually, (like you must have been, if you were manually picking the network).
[00:47:25] <killfill> in PROV. POOL it says 'excluded'.. :S
[00:48:08] <ryancnelson> yep.  that's your problem.  the provisioner won't provision on the headnode (or any other node that's excluded) unless you explicitly say "put it THERE"
[00:48:40] <ryancnelson> so edit your compute nodes… and mark it as non-reserved
[00:48:46] <killfill> greate got it. :)
[00:49:02] <ryancnelson> http://ryan.net/misc/skitch/Operations_Portal-20120706-154855.png
[00:50:06] <killfill> hm... i think i have a differe version, but anyway, what i see here is "Included in Automatic Server Selection" siwthced that and voilá
[00:50:23] <ryancnelson> yeah, the label in the portal probably changed
[00:52:10] *** deirdres has quit IRC
[00:53:04] <killfill> having an api to manage things is just avawesome
[00:53:32] *** dis has joined #smartos
[00:58:07] *** ira has joined #smartos
[01:12:02] *** jdavis has quit IRC
[01:22:32] *** deirdres has joined #smartos
[01:34:56] *** jamesd_ has quit IRC
[01:35:35] *** jamesd_ has joined #smartos
[01:43:31] *** enmand has quit IRC
[02:07:30] *** knweiss has quit IRC
[02:14:43] *** dis has quit IRC
[02:16:38] *** dis has joined #smartos
[02:26:12] *** dis has quit IRC
[02:28:12] *** dis has joined #smartos
[02:35:13] *** ryancnelson has quit IRC
[02:40:27] *** deirdres has quit IRC
[02:49:15] *** denizr1 has joined #smartos
[02:49:19] <ismell> http://pastebin.com/6Zfbd9aQ
[02:49:23] <ismell> did I do something dub, ?
[02:49:26] <ismell> dumb
[02:51:06] *** denizr has quit IRC
[02:59:39] <wesolows> create -f takes a filename
[02:59:46] <wesolows> create without -f takes stdin
[03:00:08] <wesolows> feel free to file a bug for the unhelpful error message
[03:03:25] *** dap has quit IRC
[03:07:43] *** bixu has joined #smartos
[03:09:04] <bixu> Running SmartOS on an Ultra 40 at home and, after machine became unresponsive, had to force a reboot via power button.  Now I am waiting for machine to boot and just hear drives grinding (3-way ZFS mirror is the config).
[03:09:11] <bixu> Next steps to troubleshoot?
[03:10:30] <bixu> Hmm - just booted.  Saw a console warning about no space on device.  Investigating...
[03:11:19] <bixu> "No utmpx entry.  You must exec "login" from the lowest level "shell".
[03:13:49] *** enmand has joined #smartos
[03:22:27] *** dis has quit IRC
[03:24:33] *** dis has joined #smartos
[03:28:42] <scanf> if i have 1 drive in a zpool and i want to add a 2nd mirroring drive, whats the best way
[03:29:23] <bixu> scanf: zpool attach
[03:29:36] <bixu> look at 'man zpool' and search for 'attach'
[03:29:43] <scanf> thnks
[03:29:46] <bixu> np
[03:31:08] <scanf> thoughts on zones as file servers with NFS?
[03:31:25] <scanf> the KVM zones have incredibly slow IO
[03:31:29] <bixu> I don't think that's possible.
[03:31:45] <bdha> scanf: You can't.
[03:31:53] <bixu> I think the global zone has to do the serving.
[03:31:59] <bdha> Yes.
[03:32:22] <bixu> And you can have your zones access that via autofs.
[03:32:26] <bdha> scanf: I haven't found it to be any slower than any other virtualized I/O.
[03:32:30] <bixu> At least, that's what I'm doing.
[03:33:01] <bdha> As usual, there are plenty of tools to find the latency.
[03:33:04] <bixu> scanf: Might it help to tune your zfs virt device to play nice with your KVM machine's filesystem?
[03:33:22] <bixu> I'm thinking of matching block sizes, etc.
[03:34:14] <bixu> Also, can your workload run in SmartOS?
[03:34:27] <bixu> (Had to ask the obvious question)
[03:36:35] <scanf> im trying to make a file server VM
[03:36:56] <scanf> can i enable writeback?
[03:41:50] <bixu> scanf: Can you elaborate on that?
[03:43:00] <bixu> Are you trying to speed up the KVM filesystem?  Or something else?
[03:43:10] <bixu> Also, what type of file server are you trying to create?
[03:46:02] *** denizr1 has quit IRC
[03:50:33] <scanf> bixu: LDAP, home dirs, media
[03:50:42] <scanf> bixu: yes speeding up kvm filesystems
[03:50:48] <scanf> getting ~4MB/s with virtio
[03:51:06] <bixu> Going back to my issues from earlier, I think my cron'ed snapshot jobs were causing the disk to fill and that broke SmarOS.  The only reason Unix fails :)
[03:51:30] <bixu> What sort of clients for the home dirs?
[03:51:42] <scanf> linux and solaris
[03:52:01] <bixu> You can do that from the global zone.
[03:52:48] <bixu> From experience, I'd recommend against running zones on the same machine that is your filer.
[03:53:14] <bixu> Especially if the filer is the only place the home dirs live (no cache on the client systems).
[03:53:20] <scanf> whys that
[03:53:31] <bixu> Don't put all your eggs in one basket.
[03:53:44] <scanf> true
[03:53:50] <bixu> Distribute the risk of failure across multiple systems.
[03:53:57] <scanf> what i plan to do is have a compute node and storage node and link them with infinband
[03:54:00] <scanf> *infiniband
[03:54:01] <bixu> So that when things break, you only have one mess to clean up.
[03:54:17] <scanf> not sure best way to link the storage tho
[03:54:53] <bixu> I have to say, I've moved away from the idea of central storage except as a backup for client systems.
[03:56:14] <bixu> And this coming from someone who's built many Solaris/Linux filers.  They just pile up risk in one place if you rely on them as your first storage layer.
[03:56:58] <scanf> so what do you do?
[03:57:19] <bixu> I'd be more likely to run Crashplan on each client system and have them back up to the filer.
[03:57:32] *** Licenser has quit IRC
[03:57:38] <scanf> its annoying to have 200 different unlinked versions of your homedir
[03:57:40] <bixu> Filers like the one you mention are cheap to build and a lot faster to restore from than tape.
[03:58:44] <bixu> I'm not sure what you mean by that - I guess I don't know your environment very well (obviously).
[03:59:36] <scanf> 200 servers with the same 30 users
[03:59:45] <scanf> so everyone has 200 different homedirs
[03:59:59] <scanf> think how nice itd be to just have one global homedir for everyone
[04:00:14] <bixu> oh - i see
[04:00:33] <bixu> so these aren't workstations
[04:00:55] <bixu> and therefore the durability of the backing store for the homedirs is probably not as critical as i thought
[04:01:04] <scanf> i mean it is critical
[04:01:27] <scanf> especially since thered only be 1 for each user
[04:01:37] <bixu> one server per user?
[04:01:41] <scanf> basically i want to try a solution to this at home that will scale
[04:01:59] <scanf> no, one homedir per user that is shared across all machines
[04:02:52] <bixu> you mentioned servers before - are the machines running as servers or work environments for data in these dirs?
[04:03:21] <killfill> anyone knows what are the E5-2407 cpu's?. are they the new generation R56xx or something?..
[04:04:00] <scanf> bixu: both
[04:05:04] <bixu> scanf: And how are the users accessing these machines?
[04:05:22] <scanf> bixu: ssh usually
[04:05:46] <bixu> scanf: From what type of machine?
[04:06:08] <scanf> remote endusers, shouldnt matter
[04:06:29] <bixu> Ok - so you are talking about web hosting?
[04:06:47] <bixu> How you choose to build this stuff depends largely on your target audience.
[04:07:45] <scanf> im talking about an array of services hosted on a large number of servers that the same ~30 people have shells on each one and i want to sync all the homedirs
[04:07:46] <bixu> In your shoes, I would make very different choices for virtual hosting customers vs. students vs. developers.
[04:08:30] <scanf> the provisioning system for customer-facing stuff is already inplace
[04:09:01] <bixu> So the question I'd be asking myself right now is 'what do these 30 people expect?'
[04:09:36] <bixu> (Asking that because it's not clear to me).
[04:09:47] <scanf> to make a change to their home dir (edit something, make a file, etc) on X box and have it also be that way on Y box and Z box
[04:10:06] <bixu> Is that the system they have now?
[04:10:26] <scanf> no, thats what i want to have
[04:10:58] <scanf> right now every user's homedir on every server is seperate
[04:11:06] <bixu> Ok.
[04:12:44] <bixu> I think that if you want super durable storage for lots of users, then you should consider serving that with a filer appliance like Nexenta.
[04:13:01] <bixu> SmartOS *can* do filer stuff, but it's not built for that.
[04:15:20] <bixu> My personal choice would be to sync local dirs via a central dir on the filer to distribute things and make failure less catastrophic.
[04:15:46] <bixu> But of course that's a lot of extra work :)
[04:16:12] <killfill> would smartos run on this? http://ark.intel.com/products/27214/Intel-Xeon-Processor-5110-(4M-Cache-1_60-GHz-1066-MHz-FSB)
[04:17:09] <bixu> A compromise solution would be to use autofs/nfs on the client so that user could be told to use /net/filer.yourdomain.com/users/[username] to store things they want to keep centrally.  But you wouldn't be forcing them to rely on the filer.
[04:17:57] *** enmand has quit IRC
[04:18:17] <bixu> killfill: Not sure, but since it has VT-x I think it's a good chance.  Study more if you need to run KVM hosts.
[04:19:22] <killfill> want to buy a dell 410 or something like that for a compute node and use that as the head.. just for home-testing.. :P
[04:19:40] <killfill> by that i mean the old one.. :P
[04:20:04] <bixu> I''m running SmartOS on an old Ultra 40 with AMD chips.
[04:20:11] <bixu> Can't do KVM, but I don't need to.
[04:20:50] <killfill> kvm is not used for the zones?.. just for linux?
[04:20:58] <richlowe> Right
[04:21:31] <killfill> oooh i have amd laying around..
[04:21:41] <bixu> You don't need KVM for SmartOS zones.
[04:21:49] <bixu> Which are the cool ones anyway ;)
[04:26:50] *** deirdres has joined #smartos
[04:43:48] *** enmand has joined #smartos
[04:51:39] *** bixu has quit IRC
[05:04:43] *** dis has quit IRC
[05:06:41] *** dis has joined #smartos
[05:09:22] *** bixu has joined #smartos
[05:13:41] *** bixu has quit IRC
[05:28:05] *** Saskaloon has quit IRC
[05:50:09] <ismell> is there any documentation on creating my own joyent zone dataset based of one of the existing ones ?
[05:51:45] *** bixu has joined #smartos
[05:56:22] *** bixu has quit IRC
[06:09:38] *** enmand has quit IRC
[06:13:02] *** bixu has joined #smartos
[06:16:36] *** denizr has joined #smartos
[06:17:47] *** bixu has quit IRC
[06:25:49] *** denizr has quit IRC
[06:28:50] *** Saskaloon has joined #smartos
[06:40:10] <ismell> why are somethings in /opt/local/etc and other things in /etc ? what's the difference
[06:45:27] *** Saskaloon has quit IRC
[06:52:17] *** _jared_ has quit IRC
[06:54:52] *** reiddraper has quit IRC
[06:54:53] *** gregburd has quit IRC
[06:55:46] *** Saskaloon has joined #smartos
[06:55:48] *** bixu has joined #smartos
[06:55:49] *** bixu has quit IRC
[06:56:20] *** indutny has quit IRC
[06:56:20] *** jelmd has quit IRC
[07:01:31] *** ira has quit IRC
[07:03:13] <konobi> ismell: /opt/local is pkgsrc
[07:05:42] *** jelmd has joined #smartos
[07:06:29] *** gregburd has joined #smartos
[07:07:35] *** indutny has joined #smartos
[07:07:37] *** _jared_ has joined #smartos
[07:11:04] <ismell> meaning stuff installed using pkgsrc gets put into /opt/local ?
[07:11:46] <ismell> i just make a smartosstandard zone and i have a /etc/httpd and /opt/local/etc/httpd
[07:12:50] <konobi> not sure in that scenario, but /etc is base OS stuff... /opt/local is managed by pkgsrc
[07:13:33] <konobi> i haven't looked at the smartosstandard dataset
[07:16:54] <ismell> so if i use the basic dataset and install apache using pkgsrc i will only have /opt/local/etc/httpd ?
[07:17:16] *** bixu has joined #smartos
[07:21:42] *** bixu has quit IRC
[07:21:53] *** reiddraper has joined #smartos
[08:08:48] *** Licenser has joined #smartos
[08:21:16] *** bixu has joined #smartos
[08:25:33] *** bixu has quit IRC
[08:27:17] *** AikoM has joined #smartos
[08:42:26] *** bixu has joined #smartos
[08:46:44] *** bixu has quit IRC
[09:00:11] *** darjeeling has quit IRC
[09:03:49] *** bixu has joined #smartos
[09:08:18] *** bixu has quit IRC
[09:17:47] *** marsell has quit IRC
[09:25:21] *** bixu has joined #smartos
[09:29:25] *** bixu has quit IRC
[09:46:24] *** bixu has joined #smartos
[09:47:43] *** AikoM has quit IRC
[09:50:46] *** bixu has quit IRC
[10:05:56] *** bens1 has joined #smartos
[10:13:04] *** prasm has quit IRC
[10:16:02] *** prasm has joined #smartos
[10:29:15] *** bixu has joined #smartos
[10:33:42] *** bixu has quit IRC
[10:44:03] <konobi> ismell: TIAS
[10:55:49] *** jelmd has quit IRC
[11:04:18] *** jelmd has joined #smartos
[11:09:16] *** bens1 has quit IRC
[11:11:21] *** bixu has joined #smartos
[11:54:01] *** bixu has joined #smartos
[11:59:47] *** disharmony has joined #smartos
[12:07:47] *** d[^_^]b has quit IRC
[12:24:29] *** jamesd_ has quit IRC
[12:36:43] *** bixu has joined #smartos
[12:41:04] *** bixu has quit IRC
[13:19:07] *** enmand has joined #smartos
[13:33:00] *** enmand has quit IRC
[13:46:20] <Licenser> trentster around?
[14:02:15] *** bixu has joined #smartos
[14:06:05] <trentster> howdy mate
[14:06:37] *** bixu has quit IRC
[14:23:33] *** bixu has joined #smartos
[14:28:18] *** bixu has quit IRC
[14:37:13] <Licenser> aloa
[14:38:39] <Licenser> the bug should be fixed or at lest worked around in the next dev push :)
[14:42:06] <Licenser> trentster feel free to pull :)
[14:53:45] <trentster> ok excellent! thanks
[14:58:18] *** jelmd has quit IRC
[14:58:23] <trentster> thanks, was it tought to fix?
[15:05:37] *** bixu has joined #smartos
[15:07:40] <Licenser> well it's not fixed, it's just worked around
[15:08:04] <Licenser> I forked the http server I use (cowboy) and altered the sendfiel code to catch the solaris oddity in there
[15:08:07] *** ira has joined #smartos
[15:09:59] <trentster> its strange only seems to happen consistently on the 1 node. The node only has 1 VM and the only difference is its got a lot of ram assigned to it. could be a coincedence.
[15:10:25] *** bixu has quit IRC
[15:11:17] <Licenser> you might get the errors but don';t see them if they occure selow enough erlang prevents you having trouble
[15:11:27] <Licenser> just if they fire too fast the node goes down since erlang things 'wait that isn't good!'
[15:12:13] <trentster> when you back in DE
[15:13:32] <Licenser> Hopefully end of the month
[15:14:24] <trentster> wow thats a long time thought you were only there for a month
[15:14:38] <Licenser> Heh that was the first theory yes
[15:15:08] <Licenser> But today seems to mean 'tomorrow perhaps' and tomorrow seems to mean 'sometime next week perhaps' here
[15:17:00] *** knweiss has joined #smartos
[15:28:20] *** jelmd has joined #smartos
[15:35:41] *** jamesd_ has joined #smartos
[15:40:30] *** jelmd has quit IRC
[15:48:25] *** bixu has joined #smartos
[15:48:56] *** jelmd has joined #smartos
[15:52:37] *** bixu has quit IRC
[15:56:20] *** jelmd has quit IRC
[16:02:25] *** Licenser has quit IRC
[16:04:12] <ismell> any of you install rvm ?
[16:04:18] *** Licenser has joined #smartos
[17:03:53] *** knweiss has quit IRC
[17:14:05] *** echelog-1 has joined #smartos
[17:14:47] *** rfvizarra has joined #smartos
[17:17:27] *** rfvizarra has left #smartos
[17:18:06] *** bixu has quit IRC
[17:23:49] *** denizr has joined #smartos
[17:24:51] *** denizr has quit IRC
[17:42:16] *** rfvizarra has joined #smartos
[17:56:06] *** bixu has joined #smartos
[18:00:11] *** bixu has quit IRC
[18:05:13] *** bens1 has joined #smartos
[18:17:22] *** bixu has joined #smartos
[18:23:25] *** rfvizarra has left #smartos
[18:37:54] *** rfvizarra has joined #smartos
[18:38:21] *** rfvizarra has left #smartos
[19:00:25] *** bixu has joined #smartos
[19:04:37] *** bixu has quit IRC
[19:31:47] *** AikoM has joined #smartos
[19:42:25] *** bixu has joined #smartos
[19:46:54] *** bixu has quit IRC
[20:23:46] *** david0xffea has joined #smartos
[20:24:03] *** bens1 has quit IRC
[20:25:37] *** bixu has joined #smartos
[20:30:22] *** bixu has quit IRC
[20:46:59] *** bixu has joined #smartos
[20:51:07] *** bixu has quit IRC
[21:32:58] *** david0xffea has quit IRC
[21:35:44] *** ira has quit IRC
[21:36:18] *** jelmd has joined #smartos
[22:11:42] *** bixu has joined #smartos
[22:38:27] *** killfill has quit IRC
[22:41:06] *** deirdres has quit IRC
[22:45:09] *** killfill has joined #smartos
[22:50:37] *** killfill has quit IRC
[22:51:44] *** killfill has joined #smartos
[23:08:52] *** jamesd_ has quit IRC
[23:09:14] *** jamesd__ has joined #smartos
[23:10:07] *** bens1 has joined #smartos
[23:20:38] *** bens1 has quit IRC
[23:52:13] *** jelmd has quit IRC

   July 7, 2012  
< | 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 | >