September 5, 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 5, 2013 [00:00:50] *** oddsignals has quit IRC[00:01:46] <rmustacc> Usually I find that comes from updating -extra and rebuilding it but not properly clobbering illumos-joyent.[00:02:30] *** JT-EC_ has joined #smartos[00:02:37] *** JT-EC has quit IRC[00:02:53] <konobi> yeah, testing something now[00:06:54] *** mgls has quit IRC[00:07:04] *** JT-EC_ has quit IRC[00:07:30] *** JT-EC has joined #smartos[00:10:37] *** axonpoet has quit IRC[00:13:07] *** jim80net has quit IRC[00:22:54] *** szaydel has quit IRC[00:26:44] *** emocakes has quit IRC[00:31:18] *** backjlack has quit IRC[00:32:15] *** potatosalad has joined #smartos[00:38:04] *** khushildep has quit IRC[00:42:45] *** neophenix has quit IRC[00:43:47] *** papertigers has quit IRC[00:49:48] *** tbatchelli_m has quit IRC[01:04:44] *** deirdres has quit IRC[01:04:58] *** deirdres has joined #smartos[01:10:30] *** chorrell has quit IRC[01:17:24] *** szaydel has joined #smartos[01:19:18] *** potatosalad has quit IRC[01:22:13] *** jones_d has quit IRC[01:23:40] *** deirdres has quit IRC[01:24:22] *** jones_d has joined #smartos[01:24:54] *** pringlescan has quit IRC[01:26:47] *** groundwater has quit IRC[01:36:02] *** vizant0 has quit IRC[01:41:24] *** julianduque has quit IRC[01:42:07] *** julianduque has joined #smartos[01:46:04] *** szaydel has quit IRC[01:49:48] *** szaydel has joined #smartos[01:50:49] *** nikolam has quit IRC[01:53:30] *** nikolam has joined #smartos[02:03:49] *** papertigers has joined #smartos[02:05:16] *** sa5bke has quit IRC[02:06:25] *** gkyildirim has left #smartos[02:12:07] *** sa5bke has joined #smartos[02:17:21] *** deirdres has joined #smartos[02:27:26] *** roast_ has joined #smartos[02:27:26] *** zr0 has quit IRC[02:30:41] *** mary5030 has joined #smartos[02:33:20] <roast_> i'm pretty excited to turn up smartos on my esxi server for the first time![02:34:40] <roast_> i've been thinking of replacing my freenas install with smartos, i can use kvm and dump a esxi box, is this a common usecase for your userbase?[02:38:22] *** zr0 has joined #smartos[02:48:02] *** mary5030 has quit IRC[02:48:37] *** mary5030 has joined #smartos[02:48:43] <roast_> vmware | grub | vmware | grub, rinse and repeat![02:48:59] <roast_> vmware == bios screen[02:50:39] *** FatDarrel has quit IRC[02:53:36] <wesolows> There are some people who use SmartOS as a NAS server, though with no NFS server support in zones and only Samba (no CIFS) in zones, it's really not a design centre.[02:56:41] *** potatosalad has joined #smartos[02:58:24] <roast_> wesolows: is iscsi supported in zones, how about FC?[02:59:59] <Triskelios> SCSI target support does not work in zones[03:00:04] <Triskelios> that's iSCSI/FC/IB[03:08:32] <bixu> It seems that I can't create a zpool from within a zone (using flat files as vdevs) - is this expected?[03:10:00] *** chorrell has joined #smartos[03:11:26] *** dap has quit IRC[03:12:47] <wesolows> bixu: Yes, I think so.[03:12:58] <wesolows> roast_: We don't support FC at all, full stop.[03:13:02] <bixu> Okay. Good to know.[03:13:11] <wesolows> The iSCSI target is supported in the GZ only.[03:13:28] <bixu> roast_: If you are looking for something like that, I think you want Nexenta.[03:13:32] <wesolows> We're really a rotten storage server, no better than OmniOS and worse in a number of ways.[03:13:52] <wesolows> Yet people do it all the time. *shrug*[03:14:25] <bixu> FWIW, I've gottan CIFS and NFS running in the global zone on a smartos machine. Not best practice, but it's possible.[03:14:32] <wesolows> Yes.[03:14:39] <bixu> But really, if you want fire-and-forget filer, use Nexenta.[03:14:51] <wesolows> Yup. Of course, that's not open source and not free.[03:15:02] <bixu> wesolows: I thought there was a free version?[03:15:28] <wesolows> Maybe there is now, dunno.[03:15:42] <wesolows> Ask Nexenta. :)[03:17:17] <bixu> http://www.nexentastor.org/projects/1/wiki/CommunityEdition[03:17:26] <bixu> It's still not fully open, and limited to 18TB.[03:34:39] *** pinja has quit IRC[03:35:02] *** pinja has joined #smartos[03:40:02] <natefoo> someone on irc reported that they could only get jumbo frames < ~4081 bytes through a kvm, which i am experiencing as well.[03:40:58] <natefoo> is this a known problem or have people successfully gotten 9000 byte packes through to a kvm zone.[03:41:01] <natefoo> ?[03:47:41] *** papertigers has quit IRC[04:00:59] *** mary5030 has quit IRC[04:13:18] *** chorrell has quit IRC[04:19:44] *** pringlescan has joined #smartos[05:03:53] *** leecallen35 has joined #smartos[05:07:50] *** leecallen has quit IRC[05:26:46] *** potatosalad has quit IRC[06:00:03] *** jwk404_ has joined #smartos[06:00:51] *** pringlescan has quit IRC[06:01:44] *** julianduque has quit IRC[06:01:50] *** jwk404 has quit IRC[06:03:53] *** szaydel has quit IRC[06:05:03] *** noahmehl_ has joined #smartos[06:07:07] *** noahmehl has quit IRC[06:07:07] *** noahmehl_ is now known as noahmehl[06:10:21] *** chorrell has joined #smartos[06:13:16] *** cjones_ has joined #smartos[06:28:39] <rmustacc> natefoo: I haven't heard of anyone actually using jumbo frames with our kvm stuff, but I'm not surprised there are bugs in that.[06:29:30] <rmustacc> I'll keep that in mind as another thing to add to the testing list with the current stuff I'm looking at. Probably need a better way to tell qemu what the mtu is.[06:32:46] <Teknix> I'll have to check and see if I set any of mine up like that.. I'm pretty sure I have a few[06:41:18] *** chorrell has quit IRC[06:42:30] *** FatDarrel has joined #smartos[06:48:03] *** emocakes has joined #smartos[06:49:37] *** nefilim has quit IRC[06:52:33] *** FatDarrel has quit IRC[06:57:14] *** emocakes has quit IRC[07:07:28] *** pringlescan has joined #smartos[07:12:33] *** pringlescan has quit IRC[07:40:10] *** noahmehl has quit IRC[07:43:55] *** noahmehl has joined #smartos[07:45:52] *** wolfeidau has quit IRC[07:46:45] *** wolfeidau has joined #smartos[07:48:17] *** noahmehl_ has joined #smartos[07:48:31] *** noahmehl has quit IRC[07:48:31] *** noahmehl_ is now known as noahmehl[07:55:46] *** haydock has joined #smartos[08:08:04] *** pringlescan has joined #smartos[08:13:00] *** pringlescan has quit IRC[08:13:14] *** cjones_ has quit IRC[08:13:34] *** hadret has joined #smartos[08:16:37] *** mikl has quit IRC[08:19:52] *** mikl has joined #smartos[08:24:28] *** echelog-1` has joined #smartos[08:25:02] *** echelog-1` is now known as echelog-1[08:25:02] *** echelog-1 has quit IRC[08:26:02] *** echelog-1 has joined #smartos[08:27:50] *** hadret has quit IRC[08:28:30] *** cjones_ has joined #smartos[08:29:25] *** TMinus36 has quit IRC[08:30:48] *** echelog-1 has joined #smartos[08:31:16] *** TMinus36 has joined #smartos[08:45:13] *** cjones_ has quit IRC[08:52:51] *** deirdres has quit IRC[09:05:27] *** echelog-1 has joined #smartos[09:08:36] *** pringlescan has joined #smartos[09:13:27] *** pringlescan has quit IRC[09:17:31] *** hadret has joined #smartos[09:22:15] *** alcir has joined #smartos[09:22:48] *** texarcana has joined #smartos[09:28:00] *** mamash has joined #smartos[09:29:00] *** mamash has left #smartos[09:29:06] *** mamash has joined #smartos[09:30:02] *** mamash has left #smartos[09:30:18] *** mamash has joined #smartos[09:34:56] *** vizant0 has joined #smartos[09:40:33] *** oddsignals has joined #smartos[09:48:19] *** mariusp has joined #smartos[09:51:55] *** vizant0 has quit IRC[09:54:50] <mariusp> hi guys, how do you deal with high availability for zones and vms? it seems shared storage is pretty useless with zfs but this feels to me like limiting all zones/vms to a single physical server. Am I missing something?[10:01:09] <MerlinDMC> there is no HA for zones and VMs build into the OS ... make your apps handle the failover scenario not the hypervisor[10:02:17] *** KermitTheFragger has joined #smartos[10:03:42] <MerlinDMC> you can also use zfs incremental sends to have a nearly perfect copy of your stuff on some other node but you would have to boot it there and reroute the ip to the new node in case of a failure[10:04:09] <mariusp> MerlinDMC: I get that but (and please note Im coming from the vmware, kvm shared storage ha camp) is this not a huge overhead compared to a shared storage and balancing you vms on that?[10:04:56] <mariusp> I am thoroughly enjoying smartos and the joyent cloud but how do I explaing to folks there is no HA and they are bound to a single physical server.[10:05:13] <jperkin> it's a matter of philosophy, we believe in simple and observable constructs, others believe in more complex systems[10:05:27] <jperkin> there's little merit in trying to reconcile both camps[10:06:52] <jperkin> http://www.joyent.com/blog/magical-block-store-when-abstractions-fail-us[10:07:16] <mariusp> I am not trying to reconcile just trying to figure out how to sell to those that are more technically inclined and are used to HA everything.[10:08:43] *** ktkNA is now known as ktk[10:09:10] *** pringlescan has joined #smartos[10:13:54] *** pringlescan has quit IRC[10:13:57] *** wiedi has joined #smartos[10:15:20] *** vizant0 has joined #smartos[10:20:28] *** elijah-mbp has quit IRC[10:22:10] <konobi> jperkin: http://i.imgur.com/wCsbU.png[10:25:07] *** tat- has quit IRC[10:26:19] *** rodgort has quit IRC[10:26:24] *** oddsignals has quit IRC[10:29:49] *** rodgort has joined #smartos[10:35:17] <mariusp> jperkin: nice article, thanks[10:38:42] *** bens1 has joined #smartos[10:47:15] *** andy_js has joined #smartos[10:58:53] *** tru_tru has quit IRC[10:59:57] *** tru_tru has joined #smartos[11:09:36] *** pringlescan has joined #smartos[11:14:21] *** pringlescan has quit IRC[11:21:34] *** tat- has joined #smartos[11:32:39] *** backjlack has joined #smartos[11:43:15] *** mikl has quit IRC[12:01:26] *** bens1 has quit IRC[12:01:39] *** bens1 has joined #smartos[12:10:05] *** pringlescan has joined #smartos[12:14:48] *** pringlescan has quit IRC[12:41:06] *** wolfeidau has quit IRC[12:57:47] *** mamash has left #smartos[13:02:15] *** mariusp_ has joined #smartos[13:04:46] *** mariusp has quit IRC[13:07:09] *** W0rmDr1nk has joined #smartos[13:07:10] <W0rmDr1nk> hi[13:07:17] <W0rmDr1nk> is smartos linux kernel ?[13:08:03] <jperkin> huh? no.[13:08:28] <jperkin> where would you get that idea from? ;)[13:08:33] <W0rmDr1nk> nvm[13:08:50] <W0rmDr1nk> can I ask why not just use linux ?[13:09:02] <W0rmDr1nk> I mean what does smartOS offer - except dtrace ?[13:09:24] <W0rmDr1nk> you have openvz, zfs and kvm on linux[13:09:41] <W0rmDr1nk> I'm not really trying to be antagonistic by asking this - really just want to understand the rationale[13:10:35] *** pringlescan has joined #smartos[13:10:59] <jperkin> I'm not sure really how productive a discussion on that point would be, it's up to each individual to pick the solution which works best for them.[13:11:06] <W0rmDr1nk> sure[13:11:27] <W0rmDr1nk> I also saw that Joyent worked with 10gen to make mongodb work better on SmartOS[13:11:38] <W0rmDr1nk> but ehrm, isnt zfs and mongodb kinda like mortal enemies ?[13:12:19] <W0rmDr1nk> cos zfs scraps the page cache idea - and mongodb relies heavily on the page cache - so at best you have double buffering[13:14:44] <jperkin> I have no experience with mongo, sorry.[13:15:15] *** pringlescan has quit IRC[13:18:09] <arekinath> as a side note, I wouldn't call openvz an equivalent to zones tbh. maybe the new linux containers stuff is starting to get close, but really until very very recently that kind of stuff on linux has been dodgy hacks that have all kinds of problems. :P[13:21:00] <arekinath> and my experience with zfsonlinux has been fraught with bugs and performance problems that aren't there on solaris, too. the quality of implementation/integration matters more to me than features, though. not everyone feels the same way, and software quality is a hard thing to pin down and compare[13:21:13] <arekinath> so ymmv. :)[13:26:00] <arekinath> (and dtrace is probably worth it on its own -- it frustrates me endlessly that linus and gang care so little about instrumentation and measurement tools in linux)[13:35:51] *** szaydel has joined #smartos[13:36:16] *** mariusp_ is now known as mariusp[14:06:47] <natefoo> rmustacc: Teknix: thanks.[14:09:19] *** emocakes has joined #smartos[14:10:49] <opeth__> arekinath: well roared[14:11:05] *** pringlescan has joined #smartos[14:13:07] <Alasdairrr> W0rmDr1nk: One would ask, why do people continue to use Linux when OpenSolaris/illumos is available[14:13:26] <Alasdairrr> Zones, SMF, DTrace, ZFS, COMSTAR, Crossbow, the list goes on[14:14:42] <W0rmDr1nk> Alasdairrr, well - solaris kinda sucky TBH, I mean its ahead in some areas, but far behind in others, ZFS is really nice, if you are not using it on a computer, and Dtrace is awesome, if want to know exactly how bad a choice it was to go with solaris[14:15:16] <Alasdairrr> Troll alert[14:15:27] * Alasdairrr goes back to doing work[14:15:34] <W0rmDr1nk> ehrm, I have been working a long time with solaris ... I'm not trolling[14:15:42] *** pringlescan has quit IRC[14:16:05] <W0rmDr1nk> I have had to spend hours in dtrace mdb with kernel to figure out just how bad it is[14:16:16] <W0rmDr1nk> and that there is not much I can do about it[14:17:06] <W0rmDr1nk> zfs arc was a bad choice - it messes up some very core parts of how both the CPU works with memory and makes that some things which are standard unix like mmap will just be useless[14:17:44] <opeth__> I'm back to work myself I guess ->[14:21:15] *** Shadow_aok has joined #smartos[14:21:19] <Shadow_aok> hi guys[14:24:03] <Shadow_aok> i'd like to set up an aggregation between my server 2 nics, but the second nic is down (cable is plugged and working) and I can't find how to enable it[14:24:11] <Shadow_aok> ifconfig e1000g1 up didn't work[14:24:13] <Shadow_aok> any idea ?[14:26:21] *** pringlescan has joined #smartos[14:29:04] <Alasdairrr> ifconfig e1000g1 plumb[14:29:17] <Alasdairrr> dladm show-link[14:29:25] <Alasdairrr> dladm show-phys[14:30:05] *** pringlescan has quit IRC[14:32:33] <Shadow_aok> e1000g1 phys 1500 down[14:32:55] <Alasdairrr> ifconfig e1000g1 plumb up[14:32:56] <Alasdairrr> and recheck[14:33:01] <Alasdairrr> if it still says down, check your cable/switch[14:33:09] <Shadow_aok> ifconfig: cannot plumb e1000g1: Interface already exists[14:33:16] <Alasdairrr> then just do "ifconfig e1000g1 up"[14:33:20] <Shadow_aok> maybe if i edit my config file to define a storage ip ?[14:33:37] <Shadow_aok> done and still down[14:33:46] <Alasdairrr> if its still down, then its down[14:33:52] <Alasdairrr> assigning an IP won't help[14:34:19] <Alasdairrr> Check your switch to see if the port is administratively shut down, or in an error state (such as a spanning tree bpdu error if you've got bpdu protection enabled)[14:34:59] <Shadow_aok> damn, found the problem[14:35:10] <Shadow_aok> I forgot that I already enabled port trunking on my switch[14:35:11] <Shadow_aok> thanks[14:35:24] <Alasdairrr> :-D[14:35:24] <Alasdairrr> np[14:35:51] <Shadow_aok> do you know if I can define a .vim on my usbkey to be loaded at boot ?[14:36:05] <Shadow_aok> I'd rather be able to use the arrows in vi[14:36:24] <jperkin> the canonical way to do any customisation is with something like https://gist.github.com/jclulow/2606370[14:36:38] <jperkin> and then just dump whatever you want into the postboot script[14:36:53] <Shadow_aok> sweet, thanks[14:37:00] *** emocakes has quit IRC[14:37:14] <jperkin> though I'd strongly recommend learning hjkl, your wrists will be thankful evermore.[14:37:58] <Shadow_aok> could be a good idea after all[14:40:27] <Shadow_aok> last question, does someone already had an issue when halt/shutdown wasn't able to physically power down the server ?[14:46:33] <hadret> init 5 too?[14:54:42] <Shadow_aok> yep[14:55:20] <Shadow_aok> i thought if was a bios settings issue but i didn't find anything wrong[14:55:43] <Shadow_aok> i'll try to boot on a linux livecd to check that out[14:57:10] *** deirdres has joined #smartos[15:01:03] *** elijah-mbp has joined #smartos[15:04:11] *** deirdres has quit IRC[15:04:39] *** axonpoet has joined #smartos[15:09:13] *** deirdres has joined #smartos[15:11:30] *** potatosalad has joined #smartos[15:11:41] *** deirdres has quit IRC[15:14:44] *** Tekni has joined #smartos[15:15:25] *** xinkeT has quit IRC[15:20:00] *** pringlescan has joined #smartos[15:21:32] *** alcir has quit IRC[15:22:04] *** alcir has joined #smartos[15:24:10] *** pringlescan has quit IRC[15:24:38] <leecallen35> mariusp: are you still here?[15:24:45] <mariusp> leecallen35: sure am[15:25:16] <leecallen35> I have struggled with the same issue (HA) for months. where I have landed on it...[15:25:51] <leecallen35> the HA availabilities in eg ESXi drive up costs insanely. first, the cost of vMotion. the cost of the SAN.[15:26:14] *** mamash has joined #smartos[15:26:16] <leecallen35> the multiple/fast NICs required to compensate for the performance degradation of all disk I/O going over the network.[15:26:48] <leecallen35> similarly the performance required of the SAN - not just to compensate for the network I/O, but because ALL servers share the disk.[15:26:57] <leecallen35> and of course the SAN is a single point of failure.[15:27:39] <leecallen35> so an HA implementation is X times the cost of a non-HA solution. where X is pretty big.[15:28:23] <leecallen35> I plan to focus on making my single platform as reliable as possible. ZFS of course. RAID obviously. redundand power supplies when warranted.[15:29:01] <leecallen35> frequent snapshots with sync to a backup disk/pool. also snapshots to external backup.[15:29:40] <leecallen35> keep the static stuff (OS, programs) on separate pools from the data.[15:30:21] <mariusp> leecallen35: I understand the costs and agree that it is expensive. It just does not yet feel right to discard what every other vendor has taught the industry to do: shared storage, clusters, etc. for high availability[15:30:49] <jperkin> vendors who profit nicely from such setups[15:30:53] <mariusp> :)[15:31:03] <leecallen35> I dont think anyone is suggesting SmartOS for those big corporate data centers that have lots of virtualized windows servers[15:31:28] <mariusp> leecallen35: I have tried, the SDC and would like to see it happen someday soon :)[15:31:57] <mariusp> I think as we play with the platform a little more things will fall into place and well understand the concept a little better[15:32:00] <leecallen35> I am virtualizing Windows Server too, I am just saying, its a different market[15:32:47] <leecallen35> at present the implementation precludes the shared storage and thus the failover vMotion provides[15:33:40] <leecallen35> but as was pointed out earlier, you can come close. near-real-time sync using zfs send/recv.[15:33:55] <leecallen35> but, at present, not memory[15:35:07] <mariusp> I thing sync is the way form what I gather thus far but am wondering how quickly we can restore a zone and eventual vm's? I asume its the standard import / export of both zones and pools onto a new server?[15:35:51] <leecallen35> there is a 'vmadm send/recv' that is not yet supported, and I think not yet working[15:35:59] <leecallen35> when that is working it will simplify things considerably[15:36:35] <leecallen35> I made one attempt to migrate my zones/VMs to another box using the snapshots, but it failed, I need to spend more time on that to get it working[15:36:37] *** hadret has quit IRC[15:36:37] *** Azbruh_ has quit IRC[15:36:52] *** Azbruh has joined #smartos[15:37:03] <JT-EC> So have you tested citrix that now has it's live storage motion as well as xenmotion - now does local storage to local storage too?[15:37:48] <leecallen35> my plan for almost-ESXi level failover: two SmartOS systems, with the VMs split over the two. Each syncing their zfs contents and VMs to the other. So at any time, either box can be powered down, and the necessary zones/VMs can be started on the other box.[15:40:04] <leecallen35> JT-EC: I am completely ignorant of the Citrix product[15:41:40] *** jamesd_ has quit IRC[15:43:16] *** potatosalad has quit IRC[15:45:46] <leecallen35> last thing. the joyent answer, as you heard earlier, is to achieve your HA/redundancy at the application level.[15:46:01] <leecallen35> ie, use database sync, stateless zones/VMs, etc[15:46:42] *** emocakes has joined #smartos[15:55:34] *** potatosalad has joined #smartos[15:59:16] *** neophenix has joined #smartos[15:59:42] *** pringlescan has joined #smartos[16:11:50] <JT-EC> Which package or compoent of a build system provides the /usr/bin/json command?[16:11:56] <JT-EC> *component[16:13:34] <MerlinDMC> that one is integrated in the platform ... should be in the illumos-extra repo afaik[16:15:18] *** pringlescan has quit IRC[16:16:08] <killfill> there is a npm package calles json too :)[16:16:15] <JT-EC> Doesn't appear in the manifest[16:17:12] <jperkin> note that our json is 'jsontool' on npm, 'json' is something different[16:24:02] *** FatDarrel has joined #smartos[16:28:16] *** vizant0 has quit IRC[16:29:37] <yofuh> JT-EC: http://trentm.com/json/[16:30:49] <JT-EC> Thanks[16:32:15] <yofuh> 've been searching for it some time ago[16:32:59] <jperkin> it's in the manual page, --help output, and at the top of the script itself - can't really make it more clear than that ;)[16:33:56] <yofuh> jperkin: it takes somewhat to assume a script in /usr/bin[16:34:34] <JT-EC> Yeah, I'd got to the top of the script. Still can't see it in illumos-extra but that's irrelevant to what I wanted the info for so nevermind.[16:35:08] *** mary5030 has joined #smartos[16:36:06] *** chorrell has joined #smartos[16:36:14] *** mary5030 has quit IRC[16:36:47] *** mary5030 has joined #smartos[16:36:58] <yofuh> remember not to scroll down to json_parse=function() when looking in the script...[16:38:01] <JT-EC> Looks like jq might be what I'm after.[16:38:11] *** nefilim has joined #smartos[16:38:20] <jperkin> yeh jq is pretty nifty too[16:43:24] <yofuh> fwiw https://github.com/elzr/vim-json[16:44:26] <jperkin> neat, will grab that[16:44:59] <yofuh> do me a favor and make someone put it into /etc on the gz by default ;)[16:46:12] <jperkin> oh heh, I'm already using that one.[16:46:18] <jperkin> it looked better.[16:47:36] <yofuh> it has some limitations, used it for chef-foo a lot, really makes life easier but doesnt replace json <[16:48:06] <jperkin> the GZ vim doesn't enable syntax by default, I don't have enough asbestos available to propose changing that..[16:50:26] *** emocakes has quit IRC[16:50:44] *** gkyildirim has joined #smartos[16:52:49] *** FatDarrel has quit IRC[16:53:12] *** cjones_ has joined #smartos[16:53:17] *** FatDarrel has joined #smartos[16:59:29] *** wiedi has quit IRC[16:59:35] *** tbatchelli_m has joined #smartos[17:04:51] *** leecallen has joined #smartos[17:05:26] *** vizant0 has joined #smartos[17:05:36] *** W0rmDr1nk has quit IRC[17:06:15] *** W0rmDr1nk has joined #smartos[17:07:43] *** leecallen35 has quit IRC[17:10:18] *** FatDarrel has quit IRC[17:10:40] *** cjones_ has quit IRC[17:13:32] *** tbatchelli_m has quit IRC[17:14:52] *** mamash has left #smartos[17:15:02] *** FatDarrel has joined #smartos[17:21:00] *** alcir has quit IRC[17:28:26] *** nefilim has quit IRC[17:28:45] <elijah-mbp> ha! that's like asking for a new version of ksh.[17:29:35] <jperkin> at least a new version of ksh would fix bugs..[17:29:36] <rmustacc> I'd much rather do a ksh update.[17:30:14] <rmustacc> I was la[17:30:21] <rmustacc> *almost going to set the default vim settings once.[17:30:31] <rmustacc> But then people would be stuck with something *I* like.[17:30:41] <rmustacc> But then someone said the wrong thing.[17:30:49] <rmustacc> And everyone started asking questions.[17:30:51] <rmustacc> So it got dropped.[17:34:47] *** Shadow_aok has quit IRC[17:35:22] *** dap has joined #smartos[17:55:57] *** vizant0 has quit IRC[17:56:20] *** vizant0 has joined #smartos[17:56:26] *** tbatchelli_m has joined #smartos[17:58:40] *** haydock has joined #smartos[17:58:59] *** haydock has left #smartos[18:05:48] *** FatDarrel has quit IRC[18:07:56] *** mary5030 has quit IRC[18:09:40] *** cjones_ has joined #smartos[18:17:05] *** mary5030 has joined #smartos[18:17:13] *** FatDarrel has joined #smartos[18:18:32] *** xenol has quit IRC[18:18:38] *** xenol has joined #smartos[18:20:02] *** KermitTheFragger has quit IRC[18:24:59] <MerlinDMC> morning / evening[18:29:41] <yofuh> the later[18:30:52] *** deirdres has joined #smartos[18:33:17] *** altf2o has quit IRC[18:37:11] *** tbatchelli_m has quit IRC[18:42:14] *** texarcana has quit IRC[18:42:16] *** pringlescan has joined #smartos[18:43:04] <pringlescan> hello all, LDAP (apacheds) completely fails now with this entry in the error logs repeatedly: http://pastebin.com/WMjaf2ER any ideas?[18:43:42] *** groundwater has joined #smartos[18:45:50] *** vizant0 has quit IRC[18:46:27] *** ilovezfs has quit IRC[18:46:44] *** ilovezfs has joined #smartos[18:47:12] *** ilovezfs has quit IRC[18:47:24] *** ilovezfs has joined #smartos[18:48:47] *** texarcana has joined #smartos[18:49:03] *** mary5030 has quit IRC[18:53:59] *** jaimef has quit IRC[18:58:56] *** jaimef has joined #smartos[19:00:41] *** vizant0 has joined #smartos[19:02:09] <jesse_> pringlescan, some config has too big a value for send buffer size?[19:02:18] <jesse_> which jvm are you running?[19:02:51] *** jim80net has joined #smartos[19:04:45] <pringlescan> the latest oracle one[19:04:53] <pringlescan> I give up, ApacheDS replication wiped out my directory[19:05:17] <jesse_> maybe it's no longer illumos/smartos 'compatible'[19:05:24] <pringlescan> I'll just use OpenLDAP, apache software can't run on Solaris anyway apparently as is.[19:05:40] <pringlescan> jesse_ anything that uses one of their connection handling libraries (including some Atlassian tools) won't run on SmartOS[19:05:42] <jesse_> did you try with pkgsrc jvm?[19:05:43] *** wolfeidau has joined #smartos[19:05:46] <pringlescan> it's a bummer, I set up everything in zones[19:05:51] <pringlescan> the bug was supposed to be fixed in 7[19:06:07] <pringlescan> it's a problem with SmartOS, and a problem in Java or Apache's code too, it's a two part fix (we think)[19:06:15] <jesse_> the pkgsrc/openjdk one uses different nio libraries, afaik[19:06:30] <pringlescan> I wish I had time to try :-([19:06:46] <jesse_> or rather, oracle java has special proprietary nio implementation[19:07:13] *** FatDarrel has quit IRC[19:07:30] <rmustacc> pringlescan: Did you ever actually figure out what it was trying to set the buffer size to?[19:07:34] <jesse_> make that oracle solaris jvm has ;)[19:07:51] <pringlescan> rmustacc: no :-( i'm here with 4 vendors because all the IT staff is out for the holiday[19:08:14] <pringlescan> if you give me the command to run I'll do it before I trash these zones[19:08:32] <rmustacc> If you're going to trash the zones it doesn't matter then.[19:08:37] <jesse_> dtrace is probably the easiest way to find it?[19:08:48] <rmustacc> Yes, you just want to look at which getsockopt/setsockopt is failing.[19:08:49] <pringlescan> I guess I can keep one if we think we'll fix this[19:08:58] <rmustacc> Well, it's up to you.[19:09:08] <rmustacc> I can try to help you, but I can't drive this.[19:09:09] <pringlescan> eventually, I'd like to help because I want to keep my atlassian software on there[19:09:12] <jesse_> pringlescan, I'll probably try to fix it eventually[19:09:21] <rmustacc> I mean, we run atlassian software in zones.[19:09:22] <pringlescan> that sounds good enough for me[19:09:29] <pringlescan> it's only stash and… something else[19:09:42] <jesse_> but, for that we'd need you running it on pkgsrc openjdk[19:09:52] <pringlescan> ok… one second[19:10:18] <rmustacc> You don't even need dtrace. Just truss the process.[19:10:26] <rmustacc> While it's making those calls.[19:11:03] <jesse_> true[19:11:24] <jesse_> dtrace has become too 'hammer'[19:11:29] <jesse_> everything looks like nails=)[19:11:31] *** pringlescan has quit IRC[19:11:35] <rmustacc> Well, it'll solve the problem.[19:11:45] <rmustacc> But I personally don't want to write all the scripts to solve this problem.[19:11:50] <rmustacc> I got a lot of my own to deal with.[19:14:12] <jesse_> the size is >=MAX_UINT, at least[19:14:22] <rmustacc> Hmm?[19:14:45] <jesse_> looking at the code, the argument in java side is int (signed 32bit)[19:15:19] <jesse_> (so it shouldn't be it trying to set it to some exabytes or such nonsense)[19:16:39] <rmustacc> No, it's much more likely that what's happening is that you're trying to set the buffer size beyond the current maximum.[19:16:42] <rmustacc> And you just need to tune that up.[19:16:47] <rmustacc> via ipadm.[19:27:00] <jesse_> ok, the java end is dereference-obfuscated[19:27:22] <jesse_> without knowing all the versions of the libs or apacheds no way of knowing[19:34:33] <jesse_> rmustacc, setSendBufferSize( 64 * 1024 ) is one possibility[19:34:41] <jesse_> is that more than normally allowed?[19:36:13] <jesse_> PROTO PROPERTY PERM CURRENT PERSISTENT DEFAULT POSSIBLE[19:36:19] <jesse_> tcp send_maxbuf rw 49152 -- 49152 4096-1073741824[19:36:26] <jesse_> so it would be[19:37:10] <jesse_> basically, ipadm set-prop send_maxbuf=65536 should fix the problem...[19:37:17] <jesse_> s/basically/theoretically/[19:37:24] <rmustacc> What release are you on?[19:37:32] <jesse_> joyent_20130111T010112Z[19:37:39] <rmustacc> Okay, that's a bit too old.[19:37:42] <rmustacc> This has changed.[19:37:49] <rmustacc> Or rather Seb made the naming more obvious.[19:37:50] <jesse_> zone is... uhm... 1.8.4 maybe[19:37:58] <rmustacc> The zone doesn't matter, it's the platform.[19:38:09] <jesse_> ah, ok[19:38:55] <rmustacc> Specifically there is a max_buf property and a send_buf.[19:39:06] <rmustacc> send_buf used to be called send_maxbuf.[19:39:14] <rmustacc> The problem is that it's capped internally by maxbuf.[19:39:18] <rmustacc> However, that value is 1 MB.[19:39:21] <rmustacc> By default.[19:39:22] <jesse_> tcp recv_maxbuf rw 128000 -- 128000 2048-1073741824[19:39:25] <jesse_> tcp send_maxbuf rw 49152 -- 49152 4096-1073741824[19:39:37] <rmustacc> So if you were just trying to set it to 64k that would not cause an error at all.[19:39:37] <jesse_> joyent_20130629T040542Z[19:40:01] <rmustacc> Again, too old. ;)[19:40:11] <rmustacc> But it also doesn't matter.[19:40:18] <jesse_> your feature is probably too new for pringlescan, too=)[19:40:35] <rmustacc> Sure, probably.[19:40:52] <rmustacc> But again, regardless of the presence of illumos#3942, a send size of 64k doesn't matter.[19:40:59] <jesse_> but, theoretically, on older platforms ipadm set-prop send_maxbuf=65536 should fix the problem?[19:41:19] <rmustacc> jesse_: It shouldn't even be a problem to begin with.[19:41:53] <jesse_> orcale java may do something funny too[19:41:59] <jesse_> when it thinks it's on solaris 11[19:42:19] <rmustacc> Stranger things have happened.[19:42:38] <jesse_> orcale sounds somehow better than oracle=)[19:43:13] <jesse_> for one, oracle jvm ships with a different-than-openjdk nio implementation, afaik[19:43:39] <jesse_> uses different os api than the openjdk one[19:45:20] *** mariusp has left #smartos[19:45:23] *** mariusp has joined #smartos[19:45:36] *** potatosalad has quit IRC[19:45:50] *** W0rmDr1nk has quit IRC[19:45:53] *** mariusp has quit IRC[19:46:10] *** mariusp has joined #smartos[19:47:20] *** vizant0 has quit IRC[19:51:19] *** nikolam has quit IRC[19:51:22] <jesse_> rmustacc, uhm, https://download.joyent.com/pub/iso/ has only up to smartos-20130629T040542Z[19:52:55] <jesse_> all downloads -link at http://wiki.smartos.org/display/DOC/Download+SmartOS only points to the latest release[19:53:28] *** nikolam has joined #smartos[19:53:37] <jesse_> how is one supposed to find older releases?[19:53:58] <jesse_> or a link to a file with the timestamp instead of -latest[19:54:44] <rmustacc> They're all in manta.[19:54:56] <jesse_> (-latest is the worst kind of 'standard' the internet has come up with. Tyry to figure out what a random file with version -latest is two weeks from now. or two months from now)[19:55:09] <rmustacc> Just mls the directory.[19:55:12] <wesolows> they're actually not in manta. Only July 6 and later are.[19:55:37] <jesse_> wesolows, we're looking forward here!=)[19:55:39] <rmustacc> Sorry, I assumed jesse_ meant he was looking for the newer ones.[19:55:44] <rmustacc> Even though he said older.[19:55:48] <wesolows> oh, he said he wanted older. :)[19:55:49] <jesse_> I was looking for newer ones[19:55:57] <jesse_> I found new_est_ one[19:56:09] <jesse_> started to wonder how to find stuff in-between[19:56:11] <rmustacc> mls the directory[19:56:15] <jesse_> mls?[19:56:21] *** potatosalad has joined #smartos[19:56:24] <rmustacc> There is no autogenerated http directory listing.[19:56:28] <rmustacc> They're all in there, in manta.[19:56:51] <jesse_> yes, it's all fine and dandy, as long as I can access it with a random browser[19:57:00] <jesse_> as my osx doesn't ship with manta[19:57:11] <wesolows> you can, but there's no autogeneration of index.html[19:57:18] <jesse_> nor do the boxes I'd look up new versions with[19:57:29] <wesolows> so you have to know what you're looking for. It takes about 2 minutes to install the manta tools though.[19:57:37] <jesse_> on osx?[19:57:39] <jesse_> windows?[19:57:40] <wesolows> yep[19:57:49] <wesolows> they're just an npm package[19:57:50] <rmustacc> But you can get a directory listing.[19:58:01] <rmustacc> It's just a json blob though.[19:58:01] <jesse_> on a box you have pretty much no access or permission to install anything on?[19:58:11] <rmustacc> You can get the json blob and look at it then.[19:58:13] <rmustacc> It's not great.[19:58:14] * jesse_ is not impressed[19:58:16] <jesse_> sorry[19:58:22] <rmustacc> Manta is an object store.[19:58:26] <rmustacc> It's not http semantics.[19:58:26] <jesse_> I know[19:58:29] <wesolows> it's not a web server.[19:59:09] <jesse_> this is only a step down from sun's download thinghy that made it impossible to dl server software updates with a server[19:59:25] <jesse_> (ok, there was w3m, but you needed to install it)[19:59:29] <wesolows> Not to worry, index.html generation is on the list.[19:59:31] <rmustacc> You can wget and curl whatever file you want from manta.[19:59:41] <jesse_> wesolows, *whew*[19:59:42] <rmustacc> Or go to a full link in IE and save as.[19:59:42] <wesolows> And you absolutely do not and will not need software to download files from it.[19:59:52] <rmustacc> You can even get the directory listing today.[20:00:01] <rmustacc> It just ins't in the apache outgenerated format.[20:00:03] <wesolows> You can use any browser or curl or wget or telnet to get the files.[20:00:04] <rmustacc> *isn't[20:00:05] <jesse_> I'll take back by bitching and come back with it next year if it hasn't been implemented=)[20:00:14] <rmustacc> File RFEs.[20:00:29] <rmustacc> If it's seriously blocking your use of it.[20:00:46] <rmustacc> There are lots of trade offs with doing the autogenerated html generation.[20:00:49] <jesse_> just my script trying to install newer one[20:00:53] <rmustacc> Or the RFE is that we should just have a page that does it.[20:01:04] <rmustacc> Okay, you have wget or curl?[20:01:07] <rmustacc> And something that can parse json?[20:01:22] <rmustacc> Because scripting it is easy.[20:01:29] <jesse_> "whatever is on smartos 201301'[20:01:32] <rmustacc> The latest link is basically just a symlink.[20:06:21] <rmustacc> And how do you install a script when you can't install software. ;)[20:06:37] <jesse_> it's not installed[20:06:49] *** Trixboxer has joined #smartos[20:07:17] <Trixboxer> Hi, what API can I use/extend to create dataset or fire a shell script ?[20:07:17] <rmustacc> Sorry, just pulling apart your complaints. Because most windows shops I know are so locked down you wouldn't be able to get the script on there.[20:07:51] <rmustacc> Not that not having an apache style thing isn't useful, just that it's not so emphatically broken.[20:07:57] <rmustacc> It's all standard REST.[20:08:48] <rmustacc> jesse_: I'm not sure what it is you actually want. Something useful by humans in a web browser or something useful for you to script.[20:09:26] <jesse_> both[20:09:37] <rmustacc> Well you already have the latter.[20:09:57] <rmustacc> The former is something we want to work on.[20:10:08] <jesse_> I still have no idea how to actually list the available smartos versions[20:10:33] *** bens1 has quit IRC[20:11:02] <yofuh> you'll need leard mantis api, but i suppose it is not difficult[20:11:03] <wesolows> curl -X GET https://us-east.manta.joyent.com/Joyent_Dev/public/SmartOS[20:11:05] <rmustacc> wget https://us-east.manta.joyent.com/Joyent_Dev/public/SmartOS/[20:11:36] <wesolows> The output looks like: https://gist.github.com/wesolows/6453898[20:11:37] <jesse_> browser got redirected to the newest one[20:12:08] <rmustacc> Mine doesn't.[20:12:12] <rmustacc> I get a prompt for a download.[20:12:16] <jesse_> uh, no, it downloaded it[20:12:21] <rmustacc> And cat the file.[20:12:23] <jesse_> and returned to the older url[20:12:37] <jesse_> I just missed the download it did[20:14:09] <rmustacc> Directories in manta generate a json directory list when a get is applied to them.[20:14:39] <rmustacc> The browser doesn't understand that content type and thus can't see it.[20:14:45] <yofuh> the problem is that some browsers try to parse everything and interpret it 'somehow' (implementation defined)[20:14:56] <jesse_> safari doesn't[20:15:07] <jesse_> it's a bit too anal with images, too[20:15:39] <rmustacc> No, safari downloaded the directoy listing.[20:15:43] <rmustacc> I just did it.[20:15:50] <rmustacc> cat ~/Downloads/SmartOS[20:15:56] <jesse_> ie web developers think everything is ok when their images don't have proper content-type when they have .jpg in filename[20:15:56] <rmustacc> That ain't no iso.[20:16:08] <jesse_> rmustacc, like I said above[20:16:33] <jesse_> ie would probably have decided the contents is ascii and shown it[20:16:47] <rmustacc> We have the proper content-type there.[20:16:59] <yofuh> curl works fine or me, you might stick to that[20:17:01] <jesse_> json?[20:17:27] <jesse_> I was just trying to find the listing somehow manually with safari before[20:17:30] *** potatosalad has quit IRC[20:17:31] <rmustacc> Content-Type: application/x-json-stream; type=directory[20:17:40] <richlowe> this is a whole complicated discussion for something that seems like could be done with maybe 10 lines of javascript to give a simple index[20:18:13] <rmustacc> Yes, a human should write a script which does that as well, but the last thing I want to do is parse html when there's actually a stable parseable output.[20:18:59] <jesse_> in case of safari, I'm not sure downloading the contents is the right thing to do[20:19:00] <rmustacc> Maybe e^ipi will do that.[20:19:18] <rmustacc> jesse_: If a browser doesn't know how to deal with a content-type it will opt to download it.[20:19:32] *** mary5030 has joined #smartos[20:19:32] <jesse_> except that many start guesssing[20:19:38] <jesse_> ref. ie[20:19:49] <wesolows> richlowe: Well except that tools like mls need the current thing. I have to believe that if this were trivial it would be done already.[20:20:19] <wesolows> Most browsers let you change how content types are handled.[20:20:21] <rmustacc> wesolows: I think richlowe just means to go actually make index.html in there have everything.[20:20:23] <wesolows> Maybe this is as simple as that.[20:20:31] <wesolows> Ah, yeah, that'd be easy of course.[20:20:42] *** gkyildirim has quit IRC[20:21:08] <jesse_> wesolows, well, yes. if I'd do this more than this once?[20:21:35] <wesolows> Is there ever going to be an application/x-json-stream; type=directory that you'd rather download to a file?[20:21:46] <jesse_> I was trying the (now always never-working) system of removing one path component at a time to find something 'above'[20:21:46] <wesolows> Seems unlikely.[20:22:18] <jesse_> there's probably no x-json-stream I'd like to download or show, by default[20:22:53] <jesse_> but because safari is so user-friendly, it doesn't ask stupid questions=)[20:30:49] *** potatosalad has joined #smartos[20:37:39] *** Azbruh has quit IRC[20:38:41] *** Azbruh has joined #smartos[20:40:59] *** dysinger has joined #smartos[20:47:50] *** jaimef has quit IRC[20:50:48] *** jlove has joined #smartos[20:56:27] *** jaimef has joined #smartos[21:05:11] *** Trixboxer has quit IRC[21:05:29] <joshie> Is there a way to install the solaris assembler (as) into a zone?[21:06:25] <joshie> like a good way. I copied it from my openindiana box for now :)[21:06:57] <wesolows> no good way. Why would you want to?[21:07:27] <joshie> the oracle11g install depends on it. Im sure the other would work but there doesn't seem to be an easy way to change the options in the process...[21:07:40] <joshie> s/other/gnu as/[21:07:57] <wesolows> it invokes the assembler directly?![21:07:59] <joshie> yes[21:08:08] <wesolows> wow. nuts.[21:08:15] <jesse_> why would it be?[21:08:55] <jesse_> any JIT platform does its own code generation, why would using assembler be any different?[21:09:09] <jesse_> actually, calling the assembler is the more unix way to do it=)[21:09:16] <wesolows> it's a bloody installer[21:09:21] <joshie> ^[21:09:29] <wesolows> you don't need to generate code in an installer.[21:09:35] <jesse_> maybe it precompiles stuff, instead of JITting it?=)[21:10:11] <joshie> the path of least resistance in any case is to just throw solaris as into /usr/ccs/bin/[21:10:26] <joshie> (and additionally link ar and gmake)[21:10:57] <wesolows> realistically oracle is just not supported on our platform, so this won't be the last pain you suffer[21:11:09] <Meths> Did you need lots of X libs too?[21:11:11] <joshie> my reciepe for installing oracle11g is pretty solid and clean, aside from copying that binary[21:11:23] <joshie> Meths: no, if you do a silent install with a response file you don't need it[21:11:30] <Meths> Ah, okay[21:11:33] <jesse_> yes, but what things will be different while running it?[21:11:46] <joshie> it sounds like a fun journey to find out[21:11:56] <joshie> it's been pretty functional[21:12:50] <jesse_> the things is with databases, you don't know the important parts (that is, error recovery/ACID) until it's too late[21:13:29] <joshie> it's not a production database so my concern is lacking[21:13:45] <jesse_> ah, then, happy go lucky!=)[21:20:19] *** chorrell has quit IRC[21:36:15] *** jaimef has quit IRC[21:39:13] *** vizant0 has joined #smartos[21:41:44] *** FatDarrel has joined #smartos[21:42:58] *** jaimef has joined #smartos[21:44:48] *** jaimef has quit IRC[21:51:27] *** jaimef has joined #smartos[21:55:00] *** Licenser has quit IRC[21:56:42] *** nikolam has quit IRC[21:57:04] *** Licenser has joined #smartos[21:57:37] *** jlove has quit IRC[21:59:03] *** jlove has joined #smartos[22:09:26] *** ilovezfs_ has joined #smartos[22:10:55] *** ilovezfs has quit IRC[22:12:04] *** nikolam has joined #smartos[22:12:10] *** ilovezfs_ is now known as ilovezfs[22:13:38] *** julianduque has joined #smartos[22:20:29] *** mamash has joined #smartos[22:33:53] *** marsell_ has joined #smartos[22:37:22] *** marsell has quit IRC[22:37:22] *** marsell_ is now known as marsell[22:48:16] *** W0rmDr1nk has joined #smartos[22:48:32] *** mariusp has quit IRC[22:49:22] *** mamash has left #smartos[22:50:03] *** jaimef has quit IRC[22:55:42] *** pinja has quit IRC[22:56:59] *** jaimef has joined #smartos[22:57:57] *** chorrell has joined #smartos[23:01:30] *** pinja has joined #smartos[23:08:30] *** ktk is now known as ktkNA[23:11:13] *** jlove has quit IRC[23:14:26] *** andy_js has quit IRC[23:16:47] *** axonpoet has quit IRC[23:25:23] *** bens1 has joined #smartos[23:27:02] *** bens1 has quit IRC[23:38:19] *** groundwater has quit IRC[23:39:40] *** groundwater has joined #smartos[23:39:48] *** jlove has joined #smartos[23:42:08] <jesse_> uh, 20130629T040542Z just freezes on a HP DL360G7[23:42:21] <jesse_> after listing the SD*[23:42:32] <wesolows> maybe you can try the new test image![23:42:48] <wesolows> I just made it an hour ago, and I've never tested it.[23:43:03] <rmustacc> What do you mean by after listing the SD*?[23:43:04] <wesolows> Let me upload it and you can give it a try.[23:43:33] *** chorrell has quit IRC[23:44:12] <jesse_> sd11 is /pci@0,0/pci8086,3408@1/pci103c,3245@0/sd@8,0[23:44:19] <jesse_> is the last line[23:44:32] <jesse_> second to last, sd11 at cpqary30: target 8 lun 0[23:44:45] <wesolows> well, it may be you've actually finished all the cpqary shit, but hard to tell.[23:44:46] <jesse_> before that other sds[23:45:10] <rmustacc> jesse_: You have ipmi to that?[23:45:35] <jesse_> I have ilo access, so it should be somewhere there[23:46:04] <rmustacc> If you get on a serial console then you should be able to inject an nmi and poke around in the kernel debugger.[23:46:32] <rmustacc> Or if it is actually cpquary3 then Keith's image with a new driver from hp might help.[23:48:53] <jesse_> cpqary3 251?[23:49:40] <jesse_> and no serial console, I think[23:51:32] <wesolows> well, I'm trying to upload it to Manta, but this package is busted.[23:56:14] <wesolows> the test image is at https://us-east.manta.joyent.com/wesolows/public/smartos-platform-20130905T185914Z.tgz[23:56:46] <wesolows> Replace your existing platform image with that and see what happens! The unknown is exciting!