December 26, 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 | >
December 26, 2012 [00:43:31] <Licenser> :([02:16:47] *** jamesd has quit IRC[03:18:29] *** miine_ has joined #smartos[03:21:59] *** miine has quit IRC[03:21:59] *** miine_ is now known as miine[03:28:26] *** darjeeling has quit IRC[03:33:11] *** dysinger has quit IRC[03:40:53] *** dysinger has joined #smartos[03:56:14] *** dysinger has quit IRC[04:36:03] *** darjeeling has joined #smartos[05:09:56] *** sachinsharma has joined #smartos[05:13:05] *** Thrae has quit IRC[06:01:04] *** matthewpucc_ is now known as matthewpucc[06:10:25] *** sachinsharma has quit IRC[06:53:52] *** sachinsharma has joined #smartos[07:12:19] *** indutny has joined #smartos[08:09:58] *** cjones_ has quit IRC[08:53:33] *** Cpt-Oblivious has joined #smartos[08:54:31] *** newlix_ has joined #smartos[08:54:55] *** tdmackey has quit IRC[08:58:46] *** newlix_ has quit IRC[09:18:29] *** mgt576 has joined #smartos[09:23:27] *** texarcana has quit IRC[09:23:36] *** darjeeli_ has joined #smartos[09:24:39] *** texarcana has joined #smartos[09:27:07] *** darjeeling has quit IRC[09:41:37] *** mgt576 has quit IRC[09:58:35] *** cjones_ has joined #smartos[09:59:51] *** cjones__ has joined #smartos[09:59:52] *** cjones_ has quit IRC[10:17:56] *** dysinger has joined #smartos[10:23:36] *** darjeeling has joined #smartos[10:24:53] *** dysinger has joined #smartos[10:26:52] *** darjeeli_ has quit IRC[10:27:39] *** sachinsharma_ has joined #smartos[10:30:16] *** sachinsharma has quit IRC[10:30:16] *** sachinsharma_ is now known as sachinsharma[10:34:23] *** dysinger has joined #smartos[10:34:48] *** dysinger has joined #smartos[10:39:24] *** dysinger has quit IRC[10:50:40] *** Webhostbudd has quit IRC[10:54:52] *** newlix_ has joined #smartos[10:59:19] *** newlix_ has quit IRC[11:02:05] *** darjeeling has quit IRC[11:11:40] *** arekinath has quit IRC[11:15:25] *** arekinath has joined #smartos[11:15:26] *** arekinath has joined #smartos[11:35:39] *** dysinger has joined #smartos[11:46:49] *** Sachiru has joined #smartos[12:09:12] *** xmerlin has quit IRC[12:21:01] *** xmerlin has joined #smartos[12:24:41] *** sachinsharma has quit IRC[12:40:09] *** darjeeling has joined #smartos[13:11:46] *** Cpt-Oblivious has quit IRC[13:12:04] *** darjeeling has joined #smartos[13:14:27] *** xmerlin has quit IRC[13:44:48] *** darjeeling has quit IRC[14:37:59] <Alasdairrr> jperkin: when building gcc44 in my fake sngl chroot, it does some "ls -ld" stuff in /usr/include to generate it's fixed includes[14:38:10] <Alasdairrr> unfortunately my /usr/include is all symlinks inside which screws it up[14:38:29] <Alasdairrr> I can work around by mounting /usr/include via lofs temporarily but I wondered if you'd encountered that when building gcc, and if you solved it another way[14:41:09] <jperkin> Alasdairrr: I haven't built gcc for sngl yet[14:41:36] <jperkin> but good to know, will no doubt hit the same thing[14:52:29] *** forexample has joined #smartos[14:55:23] *** newlix_ has joined #smartos[14:57:45] *** jamesd has joined #smartos[14:59:43] *** newlix_ has quit IRC[15:07:02] <indutny> heya[15:07:07] <indutny> any dtrace guys around?[15:09:23] *** darjeeling has joined #smartos[15:18:02] *** dysinger has quit IRC[15:18:49] *** forexample has quit IRC[15:18:51] *** forexample_ has joined #smartos[15:19:01] *** dysinger has joined #smartos[15:26:56] *** dysinger has quit IRC[15:27:57] *** linuxprofessor has quit IRC[15:33:36] *** daleg has joined #smartos[15:44:53] *** tonyarkles has joined #smartos[15:47:25] *** linuxprofessor has joined #smartos[15:55:38] *** le^zeek has joined #smartos[15:55:45] *** le^zeek has left #smartos[15:56:08] *** le^zeek has joined #smartos[16:03:06] *** sachinsharma has joined #smartos[16:07:58] *** tonyarkles has quit IRC[16:11:17] *** sachinsharma has quit IRC[16:14:13] *** rcamera has joined #smartos[16:18:38] *** sachinsharma has joined #smartos[16:20:39] *** darjeeling has quit IRC[16:21:07] *** darjeeling has joined #smartos[16:21:09] *** sachinsharma has quit IRC[16:33:13] <wesolows> indutny: don't ask to ask, just ask. you never know who might be able to help you when they drop in and see your question. the one way to be sure no one can help you is to not ask your question.[16:34:33] <nahamu> wesolows: in indutny's defense, the question was indeed asked in #dtrace, but your point still stands...[16:35:18] <nahamu> 07:50 < indutny> and another thing - suppose copyin() in ustack helper tries to access memory at invalid address, in this case kernel seems to be ignoring error and skiping current frame[16:35:21] <indutny> suppose copyin() in ustack helper tries to access memory at invalid address, in this case kernel seems to be ignoring error and skiping current frame[16:35:22] <nahamu> 07:50 < indutny> can I get some stack trace or any info about where that copyin() has failed?[16:35:24] <indutny> nahamu: thanks[16:35:30] <nahamu> ah, there you are.[16:35:42] <indutny> well, I'm not there[16:35:44] <indutny> I'm sleeping[16:35:52] <indutny> just forgot to turn off my notebook's speakers[16:35:53] <indutny> ttyl[16:36:06] <nahamu> heh[16:39:55] <nahamu> wesolows: which stuff are you working on these days?[16:54:31] *** Sachiru has quit IRC[17:04:28] *** xmerlin has joined #smartos[17:06:28] <wesolows> oh, this and that[17:06:37] <wesolows> same as I have been for the last 8 years :-)[17:07:44] <wesolows> today I'll be working on an 82599 ipxe driver[17:07:51] <wesolows> the last few days it's been SMF[17:08:07] <wesolows> before that it was memory corruption[17:09:36] <linuxprofessor> i'm running a kvm vm in smartos which has centos installed. centos is used to power charon-axp to emulate an AlphaServer 400 and i'm installing openvms[17:09:41] <linuxprofessor> oh the joy 0)[17:09:44] <linuxprofessor> =)[17:10:30] <wesolows> sounds slow[17:11:29] <linuxprofessor> probably a lot faster than using a real alphaserver, hehe[17:12:08] <nahamu> "64-bit, 233 MHz processor[17:12:08] <nahamu> provides outstanding[17:12:11] <nahamu> performance for the price"[17:12:19] <nahamu> ugh, embedded newlines... sorry about that.[17:12:46] <nahamu> ">Up to 384 MB memory"[17:13:00] <linuxprofessor> it's a power house[17:13:38] <nahamu> is this for teaching a course, for fun, or to run some ancient software that some sad business still needs?[17:13:49] <linuxprofessor> just for fun[17:14:03] <linuxprofessor> havent used openvms in years[17:17:04] <nahamu> wesolows: may I ask why you need a dedicated ipxe driver for the 82599? is undi-mode not good enough, or does joyent flash ipxe onto the NICs?[17:22:33] <wesolows> we don't currently flash anything onto the NICs, and my plan does not include doing so[17:22:55] <wesolows> UNDI mode would only work if we had NICs with ROMs in them[17:23:29] <nahamu> so these are NICs that don't have a PXE ROM? how do you boot iPXE?[17:23:30] <wesolows> such things are of course available, but I much prefer open source code running in protected mode to closed source code running in real mode[17:23:54] <wesolows> from a USB key, or perhaps from a different NIC that does.[17:24:36] <nahamu> that just blew my mind a little... PXE booting from NIC A to get iPXE to "really" PXE boot off NIC B...[17:24:53] <wesolows> it's merely possible. not my plan.[17:25:04] <wesolows> plugging in A and B increases costs...[17:25:20] <nahamu> true[17:27:04] <nahamu> I guess even with a USB stick in every machine, it's an image you generate once per hardware generation[17:28:19] <wesolows> depends. it's our image; we can change it whenever we want. the catch is updating it in the field if need be[17:29:28] <nahamu> I was thinking of in the field. In general once it's headed out to the field it should be working well enough that any particular machine shouldn't need an updated iPXE image.[17:29:59] <nahamu> (but the next generation of machines might have a new NIC and thus need a new image for them)[17:30:12] <wesolows> yes, though it would seem a shame to put a 4GB or 16GB flash chip in a machine and fill it with only ipxe.[17:30:45] <nahamu> if said 4GB flash chip is cheaper than upgrading to a NIC with a PXE ROM, you still come out ahead...[17:31:17] *** tonyarkles has joined #smartos[17:32:10] <wesolows> coming out a nickel ahead is what asian CMs do. that's not good enough for an american company. if there's a $10 part in the system, it needs to find a way to do $100 worth of work.[17:34:10] <nahamu> ...and now I have even less of an idea of why you're working on that iPXE driver... well done.[17:34:12] <nahamu> :)[17:34:43] <wesolows> excellent :-)[17:35:21] <nahamu> the real puzzle before me is that I have two servers, one of which was being all crashy and stupid. it seemed to be related to something with the zpool on the disks...[17:36:00] <nahamu> but I moved the disks to server B, and server B remained fine, and after editing the MAC addresses in /usbkey/config, server B is running the VMs just fine.[17:36:17] <nahamu> I put server B's disks in server A, and now server A is no longer being crashy.[17:37:44] <wesolows> weird. got a crash dump?[17:39:59] <nahamu> there are some files in /var/crash/volatile on the pool...[17:41:36] <nahamu> 5 vmdump files... at least a gig each[17:42:52] <wesolows> well, you can unpack them with savecore[17:42:59] <wesolows> and examine them with mdb[17:43:06] <nahamu> I'm unpacking one now[17:43:13] <nahamu> mdb told me I had to do that.[17:43:46] <nahamu> (a very helpful response to being pointed to the vmdump)[17:43:52] <wesolows> indeed[17:44:19] <nahamu> there's a unix file and a vmcore file... do I point mdb at both?[17:44:29] <wesolows> use the number after each one[17:44:32] <wesolows> i.e. mdb 0[17:45:58] *** sachinsharma has joined #smartos[17:47:09] <nahamu> so all I know to do at this point is run ::stack and I see a vpanic at the top...[17:48:15] <wesolows> you're looking at what led up to that, probably the thing before panic or cmntrap[17:50:02] <nahamu> so vpanic, cmi_mca_panic, cmi_mca_trap, mcetrap, i86_mwait, cpu_idle_mwait, idle, thread_start[17:50:22] <wesolows> you were in the idle loop and took an MCE. that's pretty bad.[17:50:32] <wesolows> it's probably not a software issue.[17:50:40] <nahamu> right, it felt like a hardware problem.[17:50:53] <nahamu> but I'm baffled that it seems to have magically gone away...[17:51:01] <wesolows> there may be some dcmds for extracting information from the exception state[17:51:37] <nahamu> (what does MCE stand for?)[17:51:51] <wesolows> it probably hasn't gone away. it's possible that nothing is tickling it right now, or that it's intermittent and will come back. at least 90% of the time I think something "went away", it comes back. So much so that I've decided never again to ignore things that go away.[17:51:55] <wesolows> machine check exception[17:52:09] <wesolows> you can read about them in the intel or amd architecture manuals[17:52:17] <nahamu> oh I know the problem is still there somewhere just waiting to bite me.[17:53:49] <nahamu> I'll check the other dumps to see if they are all MCEs[17:57:43] <nahamu> so far 2/2... if you have any ide about the dcmd to decode it...[17:57:59] <wesolows> ::dcmds[17:59:56] *** xmerlin has quit IRC[18:04:56] <lundh> Hi[18:05:37] <lundh> I'm having some issues with netatalk and mdns on smartos, anyone using it and can guide through it?[18:05:58] <lundh> I am unable to get the virtual machine to show up in the finder sidebar[18:08:20] <MerlinDMC> lundh, mDNSResponder?[18:08:53] <lundh> yeah, where do I confgure it?[18:10:11] <linuxprofessor> i remember something about using avahi last time i installed netatalk[18:11:31] <lundh> seems to be mdnsresponder nowdays[18:12:17] <jesse_> uhm, if I do vmadm update for nics.*.ip (zone), does the ip change right away or after the next zone restart?[18:13:15] <lundh> but I cant find any configuration for it[18:14:11] <MerlinDMC> lundh, there is no default config i guess[18:15:46] <nahamu> 5/5 crash dumps are from "Unrecoverable Machine-Check Exception"... fun times.[18:16:13] <lundh> maybe I should give up and just start using FBSD instead... at least I know how to get that running :x[18:17:56] <linuxprofessor> that's why i bought a synology nas. was just screwing things up all the time while trying out new stuff =)[18:21:23] <lundh> not running smartos anymore?[18:22:26] <linuxprofessor> yeah, but not for storage. i never was.[18:22:40] <linuxprofessor> except vm storage ofc[18:22:47] <lundh> ok[18:22:55] <LeftWing> Is synology one of the ZFS-based NAS appliances I keep hearing about?[18:23:01] <linuxprofessor> upgraded my smartos box to 24 GB of RAM today[18:23:04] <lundh> I dont have the luxury of having two different machines[18:23:07] <linuxprofessor> LeftWing: no :([18:23:15] <LeftWing> linuxprofessor: Then it's scarcely a replacement :P[18:23:28] <linuxprofessor> true[18:23:37] <linuxprofessor> i think you can use zfs in units from thecus[18:24:27] <linuxprofessor> maybe? =)[18:25:02] <lundh> I'm having serious issues with running smartos as the file server...[18:25:14] <lundh> not easy getting the software I need running in a zone[18:25:40] <linuxprofessor> i usually say it's a bad idea to virtualise your file server =)[18:25:43] <nahamu> lundh: you might want to try omnios...[18:26:23] <lundh> why?[18:26:31] <nahamu> it's more general purpose.[18:26:44] <lundh> same issues with software not being ported[18:27:02] <nahamu> http://blog.smartcore.net.au/napp-it-zfs-appliance-setup/[18:27:18] <linuxprofessor> because you don't want the storage to be dependent on a hypervisor. but that's just me.[18:27:28] <LeftWing> I would guess that we have more packages in pkgsrc than OmniTI have in IPS; but that would just be a guess.[18:27:45] <linuxprofessor> although it would be worse to virtualise it in kvm/vmware etc[18:27:46] <nahamu> LeftWing: could definitely be the case...[18:27:50] <lundh> linuxprofessor: maybe running this as a freebsd machine is not such a bad idea after all[18:28:13] <linuxprofessor> i did run freebsd as os on my file server before i bought the synology =)[18:28:24] <lundh> I do that too[18:28:35] <LeftWing> Zones are not really a hypervisor, they're just a container.[18:28:44] <linuxprofessor> LeftWing: ofc, you're right[18:29:32] <LeftWing> The thorny issue of NFS servers aside, there shouldn't really be anything that's somehow worse or not possible in a zone.[18:29:50] <LeftWing> And the NFS server thing will eventually, one day, get fixed.[18:29:51] *** potatosalad has joined #smartos[18:31:56] <lundh> I would feel much happier if I could get netatalk and mdnsresponder working[18:31:58] <LeftWing> lundh: What's your current issue?[18:32:25] <lundh> netatalk: "The version of the server you are trying to connect to is not supported."[18:32:37] <lundh> mdnsresponder: doesnt broadcast what I need[18:32:53] <lundh> so nothing in finder or terminal[18:33:13] <LeftWing> So that error is, in my experience, Mac OS X sulking because you don't have the newest authentication type configured.[18:33:32] <lundh> I have it running on freebsd[18:33:44] <lundh> with the same config, just an older version of netatak[18:34:04] <LeftWing> as in the version on freebsd is older, or the version on smartos is older?[18:34:32] <lundh> the version on freebsd is older 2.2.x compared to 3.0.1 on smartos[18:35:11] <LeftWing> I think some netatalk stuff has changed between 2.X and 3.X[18:35:25] <LeftWing> Again, I'm basing this on having recently configured it and hitting this problem.[18:36:09] <LeftWing> Did you compile this yourself, or get it from pkgsrc?[18:36:30] *** le^zeek has quit IRC[18:37:28] <lundh> compiled it[18:38:20] <LeftWing> cool.[18:38:31] <LeftWing> So, do you have a "uams_dhx2_passwd.so" somewhere in your install?[18:39:02] <LeftWing> I used --prefix=/opt/netatalk so mine is @ /opt/netatalk/lib/netatalk/uams_dhx2_passwd.so[18:39:50] <lundh> no[18:40:26] <lundh> uams_passwd.so[18:40:59] <LeftWing> My memory is a touch hazy, but I think that recent OS X really only wants to use DHX2[18:41:18] <lundh> how do I get that one then? :)[18:41:26] <LeftWing> I'm looking for my notes, one moment. :)[18:41:34] *** xmerlin has joined #smartos[18:41:36] <lundh> :)[18:42:30] <linuxprofessor> im starting to remember why im not using vms anymore[18:42:44] <lundh> hehe[18:43:34] <lundh> if I get this fixed I still have the issue of having an ancient version of transmission as my only torrent option[18:43:58] <LeftWing> So, I built this on "sdc:sdc:base:1.7.2", which is not a 64bit dataset[18:44:06] <linuxprofessor> this obsession with licenses for EVERYTHING.[18:44:21] <lundh> what?[18:44:28] <LeftWing> Transmission-2.61 is hardly ancient.[18:45:13] <lundh> true it is a service release for 2.6 but still, no 2.7[18:46:11] <LeftWing> I'm using 2.4-ish, and to be honest, am coming up blank on additional things I need it to do.[18:46:34] <lundh> thats not the issue, banned clients are[18:46:58] <wesolows> why do you think 2.61 is "banned"?[18:47:03] <lundh> its not[18:47:12] <LeftWing> OK, so you'll be wanting the "openssl", "db4" and "libgcrypt" packages installed for netatalk.[18:47:52] <lundh> dont have db4 nor libgcrypt[18:48:14] <LeftWing> Then you'll be wanting to install them![18:48:33] <LeftWing> pkgin in db4 libgcrypt[18:48:54] <lundh> just did that :)[18:48:55] <LeftWing> And then, the configure line I apparently used was: ./configure --prefix=/opt/netatalk --with-init-style=solaris --without-pam --with-bdb=/opt/local/ --with-ssl-dir=/opt/local --with-libgcrypt[18:49:34] <lundh> I used /opt/local[18:49:49] <lundh> bad idea?[18:49:51] <LeftWing> Well, I try and avoid spraying files all over stuff[18:50:01] <lundh> sounds like a good idea[18:50:03] <LeftWing> In packages it's fine, because you can uninstall it and it will be gone[18:50:09] <LeftWing> but yeah.[18:50:24] <lundh> I'll remember that the next time[18:50:52] <lundh> how do you manage your paths?[18:51:01] *** sachinsharma has quit IRC[18:51:40] <LeftWing> Depends what it is -- I don't really need anything from netatalk in PATH.[18:51:59] <LeftWing> And to be honest, since I've been moving my stuff to SmartOS, I've mostly just been able to use binary packages.[18:52:29] <lundh> ok[18:53:00] <lundh> make, lets see what happens[18:53:54] <lundh> ld: fatal: library -lcrypto: not found[18:54:32] <LeftWing> hmm[18:54:35] <LeftWing> What dataset are you using?[18:54:53] <lundh> how do I check?[18:55:10] <LeftWing> Well, when you created the zone[18:55:15] <LeftWing> You had to pick one[18:55:25] <lundh> I know that, just didnt document it[18:55:30] <LeftWing> then..[18:55:32] <LeftWing> one sec[18:55:48] <nahamu> sudo pkgin -y in smtools; sm-summary | grep Image[18:56:01] *** newlix_ has joined #smartos[18:56:03] <linuxprofessor> it usually says which dataset youre using when logging in[18:56:19] <LeftWing> linuxprofessor: until you immediately remove /etc/motd :P[18:56:36] <linuxprofessor> no reason to. hehe[18:56:40] <lundh> 1.8.4[18:56:45] <lundh> base64[18:56:49] <LeftWing> Ah, 64-bit.[18:57:10] <LeftWing> wesolows: Advise us, CFLAGS god.[18:57:44] <nahamu> I've generally gotten away with CFLAGS=-m64 LDFLAGS=-m64[18:57:45] <wesolows> CPPFLAGS="-I/opt/local/include" LDFLAGS="-L/opt/local/lib -R/opt/local/lib" CFLAGS="-m64"[18:58:04] <wesolows> the missing libcrypto is more likely to be -L/-R problems than bitness[18:58:16] <LeftWing> wesolows: In this case it works for me on base, but people on base64 get that.[18:58:17] <wesolows> also, it wouldn't hurt to make sure it's actually there :-)[18:58:26] <LeftWing> He just installed openssl.[18:58:30] <wesolows> well, are we trying to build a 32-bit netatalk, or 64?[18:58:32] <wesolows> ok[18:58:41] <wesolows> I assume 64, since anything else will suck.[18:58:42] <lundh> 64 I suppose[18:58:59] <LeftWing> Yeah, the libraries you need aren't going to be around otherwise.[18:59:09] <lundh> exactly[18:59:11] * LeftWing hugs his 32bit zones[18:59:20] <wesolows> yeah I don't mess with 64.[18:59:25] <wesolows> waiting for multilib.[18:59:30] <LeftWing> Multilib will be awesome.[19:00:18] *** newlix_ has quit IRC[19:00:30] <LeftWing> It is remarkably serene in the office today.[19:00:41] <jesse_> world without 32bit will be more awesome=)[19:01:01] <wesolows> if this thing uses autoconf, you can (a) look in config.log to see exactly what failed, and (b) probably need to put those variables in the environment for configure; i.e., CPPFLAGS=... ./configure --<options>[19:01:19] <wesolows> you may also need to pass them to make or gmake; i.e., make CPPFLAGS=...[19:01:22] <nahamu> is the main advantage of the 64bit ones for when the zone gets >4GB RAM and an individual application actually needs to address >4GB?[19:01:27] <wesolows> you're not supposed to have to do that, but...[19:01:40] <wesolows> nahamu: correct. also, on x86, more registers.[19:01:45] <LeftWing> nahamu: Larger per-process address space is the obvious advantage of 64-bit. You also get more registers on x86.[19:02:11] <nahamu> does the increase in available registers sometimes translate into better performance?[19:02:14] <lundh> more registers and faster longint calculations[19:02:20] <wesolows> on SPARC the only advantage is a bigger address space, which makes it much less interesting and therefore used only for large applications.[19:02:31] *** dysinger has joined #smartos[19:02:31] <lundh> nahamu: pretty used to be around 8 %[19:02:48] <lundh> pretty? it[19:02:50] <lundh> it[19:03:09] <nahamu> I assume it's up to the compiler to make sure they get utilized.[19:03:24] <lundh> yes[19:03:26] *** tonyarkles has quit IRC[19:03:48] <lundh> nahamu: or you if you feel like writing asm :)[19:05:00] <LeftWing> lundh: https://gist.github.com/fc8f28fa99200c798843 # the config file I'm using with netatalk[19:05:27] <lundh> thank you[19:05:29] <LeftWing> seems to correctly be able to use the system password file.[19:05:58] <nahamu> LeftWing: do you pass through a file system with lofs for that?[19:06:11] <LeftWing> I actually NFS mount one.[19:06:25] <lundh> how does that work?[19:06:26] <nahamu> from the GZ or from a different machine?[19:06:46] <LeftWing> from the GZ right now, but in the past it was another machine.[19:07:01] <lundh> is that better then using a delegate ad I do?[19:07:17] <LeftWing> It's probably slower.[19:07:31] <nahamu> lundh: the downside of the delegate is that if you destroy the zone I think the delegated dataset would go away[19:07:41] <lundh> nahamu: that is unfortunally true[19:07:55] <LeftWing> I'm doing it this way because I collapsed a multi-machine config into one machine when I moved to the US a few months ago.[19:08:09] <lundh> ok[19:08:11] <LeftWing> And haven't really bothered to do too much shuffling.[19:08:31] <lundh> i dont get this, the configure works fine its the compile that fail[19:09:04] <LeftWing> What's happening now?[19:09:13] <lundh> same thing[19:09:14] <lundh> ld: fatal: library -lcrypto: not found[19:09:18] *** tonyarkles has joined #smartos[19:09:44] <LeftWing> ls -la /opt/local/lib/libcrypto.so[19:10:04] <lundh> lrwxrwxrwx 1 root root 18 31 jul 23.29 /opt/local/lib/libcrypto.so -> libcrypto.so.0.9.8[19:10:12] <LeftWing> hmm[19:10:13] <lundh> wait a minute, why 777?[19:10:19] <LeftWing> links have no permissions.[19:10:29] <lundh> oh?[19:11:32] <LeftWing> If you can read the directory you can read the link, and then what's important is the permissions on the target of the link.[19:11:54] <lundh> yeah, ok :)[19:12:07] <lundh> just the delete part that I'm worried about[19:12:15] <lundh> but anyway, not the issue at hand[19:12:32] <LeftWing> To delete a link you need write permissions on the directory it's in.[19:13:11] <lundh> thats true, modfying the link[19:13:17] <LeftWing> So you re-ran configure with wesolows' environment suggestion?[19:14:30] <lundh> missed that, trying now[19:15:44] <lundh> same problem[19:16:12] <LeftWing> and if you try running make with the environment changes as well?[19:16:27] <LeftWing> (Which you ought not to have to do, but their autoconf stuff may be busted)[19:16:34] <lundh> thats where it fail. the ./configure... always work[19:16:56] <LeftWing> Let me spin up a zone and try.[19:17:07] <lundh> hey! nice[19:17:42] <lundh> I was just gonna way that FreeBSD is looking more and more like home but with this kind if support i might just give it a few more hours[19:18:43] <LeftWing> Well, it *is* Christmas. :P[19:18:49] <lundh> :)[19:22:23] <wesolows> ok, you beat me to it because I was busy taking delivery of $2000 worth of booze. almost always easier to just try it than to do Q&A debugging[19:22:35] <LeftWing> Indeed.[19:22:51] <LeftWing> Also, you buy the *whole* *week* worth of booze in advance?[19:23:14] <wesolows> you jest... much of it is for this weekend's vacation with 13 other people.[19:23:16] <LeftWing> :)[19:23:37] <lundh> $2000 booze in US, thats like 200 sixpacks of Brboun County Stout[19:24:15] <wesolows> stout has a very high weight to alcohol ratio, making it unsuitable for transporting into the mountains[19:24:28] <wesolows> we'd need 2 or 3 cars just for that much stout :-)[19:25:01] *** rcamera has quit IRC[19:25:05] <lundh> just buy Oskar Blues Ten Fidy. Bottle ~1+% and AMAZING[19:25:16] <lundh> ~10%[19:25:27] *** rcamera has joined #smartos[19:26:20] <lundh> eh, not bottle. can[19:27:18] <wesolows> that's more like it.[19:28:24] <lundh> G'night from the same brewery is pretty good too[19:28:50] <lundh> also canned[19:30:33] <wesolows> all the kids love cans again these days.[19:30:40] <wesolows> 21A is canning their beer too.[19:30:41] <killfill> hi[19:31:09] <lundh> 21A was a new name, never heard of them[19:32:27] <lundh> Founders Imperial Stout is on cans as well[19:33:06] <lundh> I belive[19:34:32] <lundh> any idea what might caues my problems yet? :/ Still feel inclined to return to my safe place :/[19:35:03] <lundh> I could actually do both, run FBSD in a VM and keep the data shared on NFS from the GZ[19:36:01] <killfill> the d/s in vfsstat indicates that smartos is shaping the io access, right?[19:36:05] <LeftWing> So on "base64 1.8.1" I was able to build netatalk fine.[19:36:10] <LeftWing> You're on 1.8.4, right?[19:36:13] <lundh> ye[19:36:14] <lundh> s[19:36:21] <LeftWing> I'll see if I can spin that up.[19:36:56] *** xmerlin has quit IRC[19:37:12] <wesolows> killfill: correct, that means the zone is being throttled[19:37:26] <wesolows> or, more specifically, delayed[19:38:48] <LeftWing> lundh: on 1.8.1 this was enough -- https://gist.github.com/16c40e1109850dfcb94d[19:40:32] <lundh> I'll fire up a new machine hat I havent screwed with :)[19:44:41] <lundh> can I add a lofs after creating a VM or do I have to do that when I create it?[19:45:27] <wesolows> I don't think it's possible to lofs mount across a VM boundary[19:45:34] <wesolows> at least, I don't know how that would work[19:45:45] <lundh> zone, sorry[19:46:48] *** thalin has quit IRC[19:47:26] <wesolows> looks like it should be possible via vmadm update[19:47:52] <wesolows> you'll probably have to reboot the zone[19:48:02] <lundh> good, then I cont have to rebuild this zone when I'm done experimenting[19:48:18] <lundh> nope, same issue in the clean image[19:48:53] <LeftWing> I'm part way through a build on 1.8.4[19:49:27] <lundh> ok[19:52:46] <LeftWing> hmm, that's exciting, it breaks for me too.[19:53:59] <lundh> repeatability, I like that if not the outcome[19:55:29] <nahamu> on the bright side, if you were to switch to 1.8.1 for your zone, you'll probably be able to repeat the successful build...[19:55:48] <nahamu> do 1.8.1 and 1.8.4 use the same pkgsrc repo?[19:57:29] <wesolows> things that can be reproduced can be debugged[19:57:31] <wesolows> and fixed[19:57:53] <LeftWing> I swear "V=1 make" has different results on 1.8.1 and 1.8.4[19:58:05] <lundh> whats that?[19:58:39] <LeftWing> It disables the useless, undebuggable quiet build nonsense[19:58:53] <wesolows> V=1 is the thing you have to do to eliminate the world's dumbest invention, silent builds[19:59:15] <lundh> ok[19:59:31] <wesolows> the only reason to ever have silent builds is if you're building on serial consoles, which is so incredibly rare as to be not worth thinking about.[20:01:08] <lundh> hehe[20:04:18] <lundh> any recommendations?[20:04:31] <LeftWing> Not yet. Give me a few more minutes.[20:04:45] <lundh> absoluly[20:04:51] <lundh> thats not spelled right[20:15:26] *** rcamera has quit IRC[20:32:18] *** xinkeT has joined #smartos[20:34:42] *** Tekni has quit IRC[20:36:10] <lundh> anything? :/[20:36:18] <LeftWing> Things are pretty weird in 1.8.4[20:36:30] <LeftWing> If you want to get going quickly, I would suggest 1.8.1[20:36:57] <LeftWing> I will keep looking at this, though, as it should really just work.[20:37:34] <lundh> whats weird?[20:38:10] <LeftWing> Well, I would not personally expect there to have been significant change in what configure detects between 1.8.1 to 1.8.4, but perhaps my expectations are off.[20:41:05] <nahamu> LeftWing: do they at least point at the same pkgsrc repo?[20:41:40] <nahamu> might be interesting to do a full-upgrade on a 1.8.1 before building to see if that breaks things.[20:41:54] <nahamu> if not, could be interesting to "diff" the base filesystems[20:42:29] <konobi> zfs diff to the rescue[20:43:04] <LeftWing> They appear to point at the same pkgsrc repo -- which is what I would expect from two micro releases.[20:43:21] <jesse_> panicked the last time I used it. pool version was 14 and apparently I did something 'wrong'=)[20:52:02] <LeftWing> hmm[20:52:44] <lundh> whats going on?[20:53:15] <LeftWing> I don't know, but it's possibly not dataset related after all. Have to run a few more tests.[20:53:28] <lundh> how could it not be?[20:53:44] <LeftWing> Because I was not being especially scientific, it seems.[20:53:57] <lundh> what do you mean?[20:54:38] <LeftWing> We used to deliver compilation symlinks for libcrypto in the platform, and we now do not. I did my base64-1.8.1 test on a machine with a much older platform, which still has libcrypto.so in /usr/lib. My base64-1.8.4 test was on a machine where that symlink is now only in /opt/local/lib[20:54:52] <LeftWing> And it looks like netatalk is a bit stupid when constructing its library search path.[20:54:59] <LeftWing> (And I was a bit stupid not to notice.)[20:55:28] <lundh> ahh[21:18:03] *** Cpt-Oblivious has joined #smartos[21:20:25] <lundh> I'm having no luck, today... timed out waiting for /var/svc/provisioning to move for 090673a9-740e-4111-9af4-b51cb71e3d7e[21:20:42] <lundh> is what I get when I try to create a zone with a lofs[21:27:20] <lundh> Correction: I get that whenever I try to create any new zone[21:31:22] <linuxprofessor> i managed to get rid of that error[21:33:13] <lundh> sounds like you found another?[21:34:54] <linuxprofessor> nope[21:35:03] <linuxprofessor> lemme see your json file[21:36:46] <lundh> good pastebin?[21:37:25] <lundh> http://pastebin.com/ez8mMia7 normal one, times out[21:37:54] <lundh> http://pastebin.com/hHeYmPgc lofs one, times out too[21:38:00] <linuxprofessor> in my zones i use "max_physical_memory": instead of "ram":[21:38:34] <lundh> I belive those are the same for zones[21:38:43] <linuxprofessor> oh, ok[21:38:59] <linuxprofessor> when i switched from max phys etc. to ram i managed to get a kvm vm going[21:39:06] <linuxprofessor> before that it timed out[21:39:13] <linuxprofessor> thought maybe it was the other way around here[21:39:41] <lundh> unfortunally not, the first one I pasted worked the first time around[21:39:51] <linuxprofessor> also, does it make a difference if you set the gateway in the network prefs?[21:40:04] <lundh> then I tried the second one, the one with the lofs and after that noting works[21:40:40] <lundh> yes, there is no gateway set unless you specify it[21:40:43] <linuxprofessor> have you found anything in the logs? because i didnt manage to get any good info from them[21:40:55] <lundh> (except if you use dhcp in which case the dhcp server will set one for you)[21:41:13] <lundh> which logs log this?[21:43:15] <lundh> I dont understand the zone_bh.*-logs[21:44:12] <linuxprofessor> same here, never figured it out[21:44:16] <lundh> oh, the /zones/storage have child file systems too[21:45:03] <lundh> but that should not interfer with anything if i create a vm without lofs if it is an issue at all[21:45:42] <lundh> linuxprofessor: Swede by any chanse by the way?[21:46:03] <linuxprofessor> yupp[21:46:37] <lundh> found your name on swec while googeling smartos[21:46:48] <linuxprofessor> ah, hehe[21:47:04] <linuxprofessor> i've been known to write a post or two on swec ;)[21:47:30] <lundh> I used to, even wrote a newspost or two several years ago[21:47:52] <linuxprofessor> usually the discussions are a bit. well. not nerdy enough. :-)[21:48:09] <lundh> but my home was always the now almost defuct 64bits[21:48:36] <linuxprofessor> i've always been a big irc kind of guy[21:48:43] <lundh> yeah, that was always the issue with swec[21:49:17] <lundh> me too[21:50:16] <lundh> I wonder if FreeBSD might be the better choise after all[21:50:40] <lundh> I really really like Illumos though[21:51:00] <linuxprofessor> not an easy choice[21:53:08] <lundh> not at all[21:53:49] <lundh> especially since things like vnics in fbsd are far from simple[21:54:32] *** jamesd has quit IRC[21:54:46] <wesolows> there are service logs in the child zone that may contain useful information[21:55:04] <wesolows> /zones/<uuid>/root/var/log/* and .../var/adm/* and .../var/svc/log/*[21:55:26] <wesolows> you can also use svcs -xZ to see service failures in child zones from the GZ, which may help pin down why provisioning timed out[21:56:04] <wesolows> there are undoubtedly other useful pieces of data, but I'm no provisioner expert, and they're probably on vacation. Since it is the week between christmas and new year's...[21:56:40] <wesolows> there's a support staff for JPC who would help if you couldn't provision there, but on SmartOS it's purely a volunteer effort.[21:57:15] <lundh> svcs: Pattern 'd8a9aa6b-9ba0-45db-a7ec-9c2e295c360e' doesn't match any instances[21:57:24] <lundh> and none of the logs say anyting[21:57:33] <lundh> one had a row about sshd terminating[21:57:36] <linuxprofessor> problem with the initial dataset?[21:57:46] <wesolows> hmm[21:58:06] <wesolows> svcs -xZ gave you that?![21:58:57] <lundh> svcs -xZ without any other arguments gave me nothing[21:59:07] <wesolows> ok. then there are no services in the maintenance state[21:59:09] <wesolows> that's good actually.[21:59:18] <wesolows> can you zlogin to your zone?[21:59:25] <lundh> no, status is failed[21:59:36] <wesolows> but can you zlogin to it?[21:59:46] <lundh> no[21:59:50] <wesolows> hmm[21:59:59] <lundh> zlogin: login allowed only to running zones (d8a9aa6b-9ba0-45db-a7ec-9c2e295c360e is 'installed').[22:00:12] <lundh> but vmadm list have it as failed[22:00:39] <wesolows> ok, so it's not running. which means you'll need to see why it couldn't start up, which means looking at the service logs probably.[22:00:59] <wesolows> I wish anyone other than me were around to look at this; I probably know the least about this of anyone at Joyent.[22:01:54] <lundh> hehe[22:02:17] <wesolows> if your machine can't boot, I'm your man[22:02:20] <lundh> you are all extremly helpful to us non paying users anyway :)[22:02:25] <wesolows> the provisioner... not so much[22:02:28] <lundh> hehe[22:02:59] <wesolows> still, it's just software.[22:03:47] <lundh> complicated magical software ;)[22:03:55] <wesolows> if I can reproduce it, it should be easy to track down[22:04:01] <lundh> and a few hw extensions[22:04:09] <wesolows> unfortunately, it looks like I can create 1.8.4 zones all day long[22:04:31] <lundh> what caused the issue was this:[22:04:49] <linuxprofessor> im trying to recreate the problem i had earlier today[22:04:53] <lundh> http://pastebin.com/hHeYmPgc[22:05:01] <lundh> I used that json-file[22:05:25] <lundh> /zones/storage contain three child file systems: backup, timemachine and media[22:05:48] <linuxprofessor> now i cant produce the error.[22:05:49] <wesolows> but now nothing provisions, right? even json defs that have worked before.[22:05:50] <linuxprofessor> bah[22:06:02] <lundh> actually, I had "options": "rw" under target in the filesystem-area the first time around[22:06:09] <lundh> exactly wesolows[22:09:07] <wesolows> well I can try something similar and see if it fucks up for me[22:09:20] <wesolows> in the middle of fixing a different bug at the moment though[22:09:31] <lundh> :)[22:10:56] <wesolows> wait, why do you have 2 NICs on the same tag, one DHCP and one not?[22:12:00] <LeftWing> lundh: https://gist.github.com/16c40e1109850dfcb94d[22:12:35] <LeftWing> I believe that will deal with the busted-arse configure.[22:12:49] <lundh> wesolows: one with a vlan that the switch will pass through only to my LAN one with dhcp that get an IP assinged from my ISP[22:13:07] <linuxprofessor> are you building a firewall?[22:13:24] <konobi> i need to get off my ass and get part 2 of my entry done[22:13:43] <wesolows> lundh: ok. I won't be able to test that part.[22:13:59] <wesolows> downloading the base64 dataset now; I don't normally have those around.[22:14:08] <nahamu> konobi: quit being lazy and write it! :-P[22:14:27] <konobi> we'll see what the missus says =0)[22:14:45] <lundh> linuxprofessor: no but separate internal and external traffic. this allows me to run the server with different internal and external services without involving my router[22:15:08] <linuxprofessor> ah, ok[22:15:39] <lundh> but... it requires two ethernet cables from my switch to my wireless router[22:16:02] <lundh> one from the lan-side and one from the wan-side. usless router that doesnt support vlans ;)[22:16:12] <linuxprofessor> get a better one[22:16:26] <lundh> none that I have found support what I'm doing[22:16:38] <linuxprofessor> build one yourself then =)[22:16:45] <konobi> ram isn't valid for zones, it's max_physical_memory... you also need swap[22:17:06] <konobi> if you have more than one nic, you need to mark one as primary[22:17:24] <lundh> linuxprofessor: I could route everything through this server and let tha wla-router act only as an access point[22:17:51] <linuxprofessor> that's what im doing[22:17:56] <linuxprofessor> using pfsense[22:18:18] <lundh> konobi: I thought ram and max_phys was the same in zones?[22:18:33] <lundh> I'll change it[22:18:35] <konobi> nope[22:18:41] <wesolows> I cannot reproduce this problem with lofs.[22:18:42] <konobi> well, maybe at create time[22:19:37] <lundh> wesolows: could it have been the external dhcp that timed out?[22:19:42] <wesolows> a base64 zone with the following was successfully created, and I can see the contents of the lofs mount logged into the zone: https://gist.github.com/711743649f0a51cd8fa9[22:19:54] <konobi> lundh: `sysinfo | grep Zpool` ?[22:19:59] <wesolows> I don't know. Hopefully once you fix the things konobi is pointing out, things will work better.[22:20:48] <konobi> wesolows: just thinking there may be some of the SYSPOOL stuff missing[22:21:27] <lundh> how do I mark it as primary?[22:21:38] <konobi> primary: true[22:21:56] <lundh> thank you[22:22:53] <lundh> konobi:[22:22:54] <lundh> "Zpool": "zones",[22:22:54] <lundh> "Zpool Disks": "c1t0d0,c1t2d0,c1t5d0",[22:22:54] <lundh> "Zpool Profile": "mirror",[22:22:54] <lundh> "Zpool Size in GiB": 3654,[22:23:14] <lundh> not a 3-way mirror. c1t0d0 is a l2arc[22:24:14] <konobi> lundh: got ssh into that machine?[22:24:27] <lundh> yes[22:24:45] <konobi> mind if i hop on and look, might be easier and quicker[22:25:16] <lundh> absolutly. its a fresh install[22:25:48] <lundh> give me 2 sec[22:28:41] <lundh> ok, so with the changes and an unexpected reboot (dont ask) it works[22:28:50] <lundh> the changes to the json that is[22:32:30] <konobi> yeah, looks good to me[22:33:27] <lundh> so if I can get netatalk to install I might still have a shot at getting this to work[22:33:42] <konobi> only one vnic though[22:33:54] *** wolfeidau has quit IRC[22:34:05] <lundh> yesh, removed the other one. didnt want to spam my ISP with ip-requests[22:34:19] <lundh> that might have been what caused the timeout. I wount be suprised[22:34:24] <lundh> we can try if you want to[22:34:32] <konobi> https://opusmagnus.wordpress.com/2012/11/01/quick-tip-netatalk-3-0-1-on-joyent-smartos/[22:34:49] <lundh> I know, that guide doesnt work :)[22:34:55] <konobi> huh, odd[22:35:15] <lundh> he removed to much stuff so I cant actually connect to the netatak share with newer versions of osx[22:35:42] <lundh> LeftWing: didnt realised that you had a solution. yay![22:36:01] <LeftWing> lundh: does it build for you?[22:36:15] <lundh> getting everything ready to try[22:36:20] <LeftWing> cool[22:37:01] <lundh> had a few issues...[22:37:17] <lundh> while messing with other things[22:40:00] *** Teknix has quit IRC[22:42:02] <lundh> I belive LeftWing might be the netatalk hero of the day[22:42:07] <konobi> time to upvote this... https://github.com/joyent/smartos-live/issues/153[22:45:43] <wesolows> some sort of... DTD, perhaps...[22:45:55] <nahamu> just built a fresh spice dataset against the latest spice server code (which conveniently enough is the latest stable release)[22:46:05] <konobi> i think magnus must have been doing it on a base image[22:46:33] <lundh> how do I create a user in a zone the smartes?[22:46:42] <lundh> I have beeen doing it all manually so far[22:47:05] <nahamu> lundh: at zone creation time, or do you mean with the useradd command?[22:47:13] <lundh> either way[22:47:25] <lundh> or both, didnt know I could at creation time[22:47:33] <konobi> just useradd[22:47:46] <nahamu> I'm not sure you can, come to think of it. that might be a crazy hack I added to one of my zone templates...[22:48:08] <konobi> user_script[22:50:50] <lundh> yes LeftWing. I can even connect![22:51:19] <konobi> wesolows: yeah, jsonschema or rx would work well[22:51:55] <konobi> can even generate docs (and error messages) from them[22:55:30] <LeftWing> lundh: excellent![22:56:07] <linuxprofessor> lundh: now, go compile rtorrent! :)[22:56:18] <konobi> pppffftt[22:56:24] <lundh> linuxprofessor: I looked into that. way way to many dependencies[22:56:25] * LeftWing hugs transmission[22:56:43] <lundh> I'm migrating to transmission[22:56:45] <konobi> sabnzbd+ ++[22:56:53] <linuxprofessor> i guess i COULD have a look at transmission[22:57:11] <linuxprofessor> but i really like rtorrent + rutorrent[22:57:20] <lundh> LeftWing: not sure why:[22:57:20] <lundh> [Homes][22:57:21] <lundh> basedir regex = /home[22:57:27] <lundh> doesnt work though[22:57:49] <nahamu> did the user you created get /home/<username> as the home directory?[22:58:32] <lundh> yes[22:59:04] <LeftWing> Well, hopefully all problems from now on are at least not SmartOS-specific. :P[22:59:15] <lundh> nope :)[22:59:54] <lundh> got the home dir working :)[22:59:58] <LeftWing> cool[22:59:58] <lundh> but not zeroconf[23:00:19] <konobi> you can just enable mdns[23:00:21] <LeftWing> Unfortunately I know next to nothing about multicast dns.[23:00:54] <lundh> konobi: oh is that enough to tell everyone that I am a afp server too?[23:01:04] <lundh> cause finder have to know that[23:01:42] <konobi> lundh: not sure[23:02:08] <konobi> you can tell it to announce for you iirc[23:02:24] <nahamu> did you try just "svcadm enable dns/multicast"[23:02:26] <lundh> anyway. got it working by installing mDNSResponder[23:03:05] <lundh> multicast is not enough for finder[23:04:05] <konobi> dns-sd[23:04:44] <konobi> http://cafenate.wordpress.com/2009/02/26/making-netatalk-discoverable-in-opensolaris/[23:05:10] *** wolfeidau has joined #smartos[23:05:15] <lundh> look at that :)[23:05:30] <konobi> lundh: we expect wiki docs![23:05:58] <lundh> You will get it. after all this work definatly[23:10:16] *** gkyildirim has joined #smartos[23:10:39] <nahamu> hi gkyildirim[23:11:59] <lundh> /var/svc or /lib/svc?[23:12:38] <LeftWing> You shouldn't (and can't) put things in /lib/svc.[23:12:42] <lundh> ok[23:14:47] <lundh> thins that go in /lib/svc/method in solaris sould be in /var/svc/method in smartos?[23:15:42] <LeftWing> I would probably just put things in /opt/wherever/bin[23:15:47] <LeftWing> (script things, that is)[23:15:47] <lundh> oki[23:16:09] <LeftWing> /var/svc/manifest/... is the correct place for smf manifests, though (the xml files)[23:16:20] <lundh> yes[23:16:32] <LeftWing> Although technically you can put those anywhere as well, as long as you import them afterwards.[23:16:47] <konobi> on teh GZ[23:16:56] <konobi> in a zone, you can just import once[23:17:07] <LeftWing> Right, the GZ is a different kettle of fish.[23:17:21] <lundh> this is in a zone :)[23:17:38] <konobi> svccfg import < foo.xml[23:18:43] <lundh> not svccfg -v import ?[23:19:05] <konobi> either way[23:19:23] <lundh> ok[23:21:08] <lundh> Aaaand it works![23:21:21] *** wolfeida_ has joined #smartos[23:23:41] <lundh> but now I get two...[23:23:45] <lundh> this is weird[23:24:12] *** wolfeidau has quit IRC[23:24:13] <konobi> kill off mdnsresponder[23:24:23] <lundh> I have[23:24:43] <konobi> huh[23:24:45] <lundh> even done: pkgin remove mDNSResponder-258.14[23:26:10] <lundh> its netatalk that has it built in[23:26:35] <lundh> it must have gotten the feature since I had mdnsresponder installed while configure[23:27:18] <lundh> zeroconf = no solved that[23:29:04] *** jamesd has joined #smartos[23:33:28] <lundh> still cant ssh using the display name though[23:34:32] <konobi> foo.local ?[23:34:38] <lundh> didnt work[23:34:54] <konobi> from OSX?[23:34:57] <lundh> yeah[23:35:04] <konobi> huh, odd[23:35:34] <lundh> I can use the uuid though[23:35:53] <lundh> I probably should change the hostname[23:35:59] <konobi> yeah[23:39:48] <lundh> is it not enough to change through vmadm and then reboot the zone?[23:58:46] <lundh> apparently not[23:59:23] <konobi> need to edit /etc/hostname.<iface>