December 14, 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 14, 2012 [00:03:05] *** dubban has quit IRC[00:08:38] *** kevinykchan has quit IRC[00:14:57] *** jesser has joined #smartos[00:19:26] *** montyz0 has quit IRC[00:37:29] *** wdent has joined #smartos[00:45:08] *** jim80net has quit IRC[00:48:15] *** tonyarkles has joined #smartos[00:48:16] *** daleg has quit IRC[01:00:52] *** scubasteve has joined #smartos[01:09:17] *** tonyarkles has quit IRC[01:11:48] *** rancor has quit IRC[01:42:32] *** scubasteve has quit IRC[02:00:36] *** xmerlin__ has joined #smartos[02:08:33] *** dap has quit IRC[02:16:07] *** wdent has quit IRC[02:18:21] *** wdent has joined #smartos[02:25:47] *** ahaydock has quit IRC[02:26:25] *** montyz0 has joined #smartos[02:48:45] *** potatosalad has quit IRC[02:52:04] *** potatosalad has joined #smartos[03:13:02] *** rc10 has joined #smartos[03:13:43] *** montyz0 has quit IRC[03:20:38] *** potatosalad has quit IRC[03:35:49] *** potatosalad has joined #smartos[03:45:15] *** xmerlin__ has quit IRC[04:11:21] *** potatosalad has quit IRC[04:13:03] *** trentm has joined #smartos[04:16:20] *** trentm has quit IRC[04:17:05] *** wolstena1 has quit IRC[04:21:05] *** mikl has quit IRC[04:22:51] *** tonyarkles has joined #smartos[04:23:30] *** wdent has quit IRC[04:27:19] *** wdent has joined #smartos[04:35:31] *** ira has quit IRC[04:44:30] *** ryancnelson1 has quit IRC[04:46:50] *** potatosalad has joined #smartos[04:59:19] *** tonyarkles has quit IRC[05:00:20] *** potatosalad has quit IRC[05:13:13] *** ryancnelson has quit IRC[05:17:39] *** ryancnelson has joined #smartos[05:28:17] *** xinkeT has joined #smartos[05:28:27] *** Thrae has joined #smartos[05:30:52] *** Tekni has quit IRC[05:30:56] *** wdent has quit IRC[05:31:15] *** Tekni has joined #smartos[05:34:12] *** xinkeT has quit IRC[05:34:28] <Thrae> Hey, I've been using FreeBSD for a while now after the death of OpenSolaris and I'm thinking of trying out SmartOS after getting so frustrated with VirtualBox OSE Headless and a few BSD quirks. Could I tweak my existing FreeBSD 9.0 64-bit install, or would I have to rebuild world or otherwise reinstall?[05:38:27] *** e^ipi_laptop has joined #smartos[05:51:42] <LeftWing> Thrae: You mean ... you have a machine that runs FreeBSD directly on the metal, now, and you want it to run SmartOS instead?[05:53:37] <Thrae> LeftWing: Yep! I assume I'd at least need to recompile some programs and disable optimizations like march=native, but world was never recompiled from the release.[05:55:29] <Thrae> Or is there a special KVM kernel that would be needed? I've never used KVM before.[05:55:54] <LeftWing> So, FreeBSD and SmartOS are totally different operating systems with totally different ways of installing...[05:56:19] <LeftWing> We provide KVM as a kernel module that gets loaded when you fire up a VM.[05:57:12] <LeftWing> Also, everything you're presently doing on your FreeBSD machine you'll likely want to move into one or more Zones.[05:58:04] <LeftWing> It's not like you can just reboot into SmartOS, though -- the installer will want to erase your existing machine and lay down a new ZFS pool. This destroys the data.[05:59:06] <Thrae> That's the plan. Like I said, I used OpenSolaris for a long time, even spending a little time on Solaris 11 Express until I found out all the caveats, so I'm well used to the Solaris environment.[06:00:15] *** sachinsharma has joined #smartos[06:00:16] <LeftWing> cool[06:00:17] *** montyz0 has joined #smartos[06:01:21] <LeftWing> So a lot of stuff in SmartOS is a bit different from how things were in OpenSolaris.[06:01:36] <ryancnelson> don't understand your question about "could i tweak my existing freebsd 9.0 install" … you'd be erasing the whole OS[06:01:40] <arekinath> Thrae: were you asking if you could turn your freebsd install into an image to run as a VM under smartos?[06:02:01] <Thrae> FreeBSD is already using ZFS for everything, and I have two 500GB disks in a vdev stripe I'm just using for bulk data.[06:02:11] <Thrae> arekinath: Yes, exactly.[06:02:58] <arekinath> Thrae: you could probably convince the freebsd to boot up inside a KVM, sure. you'll have to take bit-for-bit images of your disks though and expose them as devices the same as they're set up now[06:03:08] <LeftWing> Ah, well you should certainly be able to run 64-bit FreeBSD as a KVM guest, provided your CPU is Intel and has VMX + EPT.[06:03:09] <arekinath> Thrae: takes up a lot of space[06:05:19] <arekinath> and running zfs in a kvm on top of zfs seems a tiny bit silly, but I've seen worse[06:05:20] <arekinath> ;)[06:06:13] *** ryancnelson has quit IRC[06:07:16] <Thrae> Yeah, now that I think about it, I could just do a zfs snapshot + send of the FreeBSD install and move the data to a simple UFS setup, keeping the snapshot as a backup in case something in SmartOS fails. No need for a zroot mirror.[06:08:21] <Thrae> I've been plenty of zfs send/receive restores. So simple and amazing.[06:08:38] <Thrae> s/been/done/[06:09:05] <arekinath> yeah. and move everything you can into zones, they're quite a bit more efficient than kvms. :)[06:09:21] <arekinath> at least that way you wouldn't need big bit-for-bit copies of your disks[06:10:36] <Thrae> Ah wait, here's a problem...KVM requires EPT, no?[06:10:54] <arekinath> yes, it does[06:13:57] <Thrae> Yeah my Q9800 supports VT-x and SLAT but not EPT.[06:15:40] *** rc101 has joined #smartos[06:15:44] <Thrae> Q9800? Q9650[06:16:14] *** rc10 has quit IRC[06:18:04] <Thrae> Hmm, can't use AMD's NPT either?[06:21:12] *** darjeeling has quit IRC[06:23:02] <LeftWing> There's experimental support for AMD-V/SVM + NPT that I can give you in a custom kvm module.[06:23:11] <LeftWing> It's certified "works on my machine". :P[06:23:37] <LeftWing> If you're happy with just zones, and don't need to run other OSes as KVM guests, then any 64-bit x86 CPU will do, of course.[06:26:45] <Thrae> Well honestly I have been planning on rebuilding my home server with real server hardware rather than repurposed gaming hardware.[06:32:46] *** rc101 has quit IRC[06:51:49] *** notmatt has joined #smartos[06:56:18] *** Plazma has left #smartos[07:03:14] *** montyz0 has quit IRC[07:05:48] <damacus> hehe[07:06:07] <damacus> yeah, it's nice having a proper server build at home.[07:06:25] *** rc10 has joined #smartos[07:07:29] *** enmand has quit IRC[07:07:40] <damacus> up until this year, i'd largely done the same as i updated my primary rig. the second-best box was my server.[07:07:55] *** Saskaloon has quit IRC[07:24:39] *** marsell has quit IRC[07:33:04] *** notmatt has quit IRC[07:50:01] *** notmatt has joined #smartos[08:02:12] *** rc10 has quit IRC[08:03:26] *** darjeeling has joined #smartos[08:16:11] *** alucardX has joined #smartos[08:18:40] *** jesterMWA has joined #smartos[08:19:47] <jesterMWA> HI,no module mdb_v8 could be found, somebody can help me. how to find it[08:24:50] *** kschiess has joined #smartos[08:26:59] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[08:30:12] *** Saskaloon has joined #smartos[08:32:12] *** rc10 has joined #smartos[08:34:55] <jesterMWA> HI,no module mdb_v8 could be found, somebody can help me. how to find it[08:36:04] <MerlinDMC> jesterMWA, most devs are sleeping now i guess[08:36:22] *** alucardX has quit IRC[08:37:56] <jesterMWA> i got it ,thanks[08:39:19] *** alcir has joined #smartos[08:53:39] *** bens1 has joined #smartos[09:10:59] *** alucardX has joined #smartos[09:23:02] *** texarcana has quit IRC[09:24:16] *** texarcana has joined #smartos[09:35:15] *** darjeeling has quit IRC[09:39:42] *** e^ipi_laptop has quit IRC[09:48:49] *** darjeeling has joined #smartos[09:50:27] <alucardX> hi[09:50:44] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[09:50:53] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[09:50:56] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[09:50:59] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[09:51:03] <jesterMWA> who can help me please[09:53:30] <trentster> Whats the current opinion on LSI 2208 and SmartOS, well supported?[09:56:29] <trentster> and what about LSI 2308 , anyone using these boards with SmartOS and can offer any feedback?[10:05:10] *** mamash has joined #smartos[10:15:07] *** rc101 has joined #smartos[10:15:08] *** rc10 has quit IRC[10:19:38] *** marsell has joined #smartos[10:24:04] *** KermitTheFragger has joined #smartos[10:24:39] <MerlinDMC> trentster, i have two LSI controllers running ... but i can't remember which one but could be in the wiki[10:25:18] <trentster> MerlinDMC: thanks[10:29:18] <jesterMWA> mdb: ld.so.1: mdb: fatal: /usr/lib/mdb/proc/v8.so: wrong ELF class: ELFCLASS32[10:30:39] <MerlinDMC> jesterMWA, repeating that does not help much i guess ... you need to wait for the devs to get up - or just create an issue on github[10:30:58] <MerlinDMC> if you're an joyent public cloud customer create a support ticket there[10:31:33] <jperkin> jesterMWA: you need to attach to the process first[10:31:52] <jperkin> by default mdb will run in 64bit mode, hence the arch mismatch[10:36:01] *** rc10 has joined #smartos[10:39:36] *** rc101 has quit IRC[10:53:34] *** dubban has joined #smartos[11:07:11] *** d[^_^]b has quit IRC[11:07:18] *** d[^_^]b has joined #smartos[11:07:47] *** d[^_^]b has quit IRC[11:07:54] *** elijah-m_ has joined #smartos[11:08:37] *** jesterMWA has quit IRC[11:09:44] *** elijah-mbp has quit IRC[11:12:18] *** d[^_^]b has joined #smartos[11:36:39] *** jamesd_ has joined #smartos[11:39:18] *** jamesd has quit IRC[11:45:17] *** darjeeling has quit IRC[11:49:37] *** riceo has joined #smartos[12:28:46] *** sachinsharma has quit IRC[12:34:05] *** bens1 has quit IRC[12:44:08] *** darjeeling has joined #smartos[13:15:23] *** izaki has quit IRC[13:26:19] *** ira has joined #smartos[13:35:38] *** bens1 has joined #smartos[13:42:30] *** jesser has quit IRC[13:43:39] *** alucardX has quit IRC[14:11:59] *** wdent has joined #smartos[14:21:29] *** wdent has quit IRC[14:22:56] *** wdent has joined #smartos[14:44:32] *** rc10 has quit IRC[14:52:03] *** enmand has joined #smartos[15:01:07] *** kamilr has joined #smartos[15:01:37] <kamilr> hi[15:01:48] <kamilr> is there any way to change file imgadm.js ?[15:02:27] <jperkin> two ways, build your own version of smartos with the change made in the source, or make a copy somewhere writeable, then mount that file on top of the original with lofs[15:03:41] <konobi> 3, submit a pull request[15:04:07] <jperkin> konobi: in this case, that's not going to happen :)[15:05:37] <kamilr> IMO there should be possibility to change the default zpool. I have searchd and i think its need to change only one file: imgadm.js which is read-only. There is zpool variable which defines default zpool[15:08:16] *** jim80net has joined #smartos[15:09:47] <konobi> there's lots of other places[15:10:49] <konobi> unless you mean as to where the images get downloaded to[15:14:42] <kamilr> yes, downloaded to.[15:18:22] <konobi> well, i'd like to see them as children under another dataset... but not a different pool[15:19:26] <kamilr> if images would be on pool zones, there would be no chance to have VM on different[15:19:41] <kamilr> so images should be on the same pool[15:21:41] <konobi> ?[15:22:26] *** ripratm has joined #smartos[15:23:29] <kamilr> If i would like to have VM on pool DATA i also will have to have images on pool DATA[15:23:48] <kamilr> i can't have images on pool zones and VM on pool data because of clones[15:26:18] <konobi> "zones" won't get changed as the place[15:27:10] <kamilr> ok, BUT if i will mount imgadm.js with changed zpool to DATA, imgadm will import images to pool data[15:27:21] <kamilr> and all VM will be created on the pool[15:28:30] <konobi> nope... there's much more to it than just imgadm[15:29:34] <kamilr> And this is the reason why i think smartos isn't as good as i thought[15:30:03] <kamilr> It is stupid that i can't use other pool for VM's[15:30:10] <jperkin> kamilr: you just need to understand that smartos has a very tight focus[15:30:25] <kamilr> jperkin: i can't[15:30:39] <jperkin> and if it doesn't work the way you want, you have the power to fork.[15:30:40] *** mamash has left #smartos[15:32:41] <kamilr> So, do you have any ideas how can i run VM's on other pool ? Is it even possible?[15:32:44] <konobi> there's an infinite number of names to use for a zpool... we picked one and went with it[15:32:56] <konobi> what's so wrong with "zones"?[15:33:21] <kamilr> local drives[15:33:31] <kamilr> i want to mount iscsi[15:33:34] <jperkin> kamilr: as with pretty much any other question about customising smartos "it is possible, but it will probably be a lot of work"[15:33:40] <konobi> not supported[15:33:48] <jperkin> it's unix and open source, after all, there's not much that _can't_ be done[15:33:51] <kamilr> what not supported ?[15:33:57] <kamilr> i have allready mounted[15:35:21] <kamilr> so i have iscsi mounted. I would like to build on it VM's. So if one physical machine goes down, i can easly mount iscsi to another[15:36:04] <kamilr> but if i have zones pool with 100gb, i'm not able to store much vm's[15:36:18] <kamilr> so i would like to have bigger pool like iscsi[15:36:26] <konobi> vms will always be local storage... there's several long discussions on the topic on the mailing list[15:36:57] <kamilr> BUT imgadm stores images on zones, and vm's must be on the same pool where images are[15:37:11] <kamilr> ok[15:37:34] <kamilr> what if local storage is not enough[15:37:35] <kamilr> ?[15:37:45] <konobi> get more diskspace[15:37:46] <jperkin> then you buy another machine[15:38:25] <kamilr> excly, buy another machine. This does not make sense.[15:38:27] <jperkin> but seriously, we hashed this out recently on the list, it's not worth repeating it all over again[15:38:30] <kschiess> kamilr: You might want to try basing yourself on OmniOS and this: http://kschiess.bitbucket.org/qemu-toolkit/README.7.html[15:38:34] <kamilr> I have big storage array[15:38:41] <kschiess> kamilr: I am the author of it, so I am biased.[15:45:47] *** montyz0 has joined #smartos[15:51:30] *** montyz0 has quit IRC[15:52:05] *** montyz0 has joined #smartos[15:52:44] *** rc10 has joined #smartos[15:54:27] *** rc101 has joined #smartos[15:57:11] *** rc10 has quit IRC[15:58:37] *** khushildep has joined #smartos[15:59:38] *** montyz0 has quit IRC[16:03:53] *** daleg has joined #smartos[16:05:55] *** neophenix has joined #smartos[16:07:55] *** wdent has quit IRC[16:08:59] *** wdent has joined #smartos[16:09:56] *** wdent has quit IRC[16:19:52] *** riceo_ has joined #smartos[16:24:12] *** riceo has quit IRC[16:25:03] *** bens1 has quit IRC[16:27:41] *** rancor has joined #smartos[16:31:29] <kamilr> omnios doesn't have such tool like imgadm ?[16:33:45] *** wdent has joined #smartos[16:34:13] *** wdent has quit IRC[16:38:22] <jesse_> imgadm is a smartos thing[16:41:42] *** tonyarkles has joined #smartos[16:47:40] <MerlinDMC> kamilr, omnios has own management frontend written in ruby ... but i can't remember the github repo[16:51:54] *** wdent has joined #smartos[16:52:32] <kamilr> you mean qemu-toolkit ?[16:54:07] *** tonyarkles has quit IRC[16:55:27] *** wdent has quit IRC[16:56:03] *** wdent has joined #smartos[16:56:13] *** tonyarkles has joined #smartos[16:58:26] *** montyz0 has joined #smartos[17:00:48] *** wdent has quit IRC[17:01:29] *** wdent has joined #smartos[17:02:04] *** montyz0 has quit IRC[17:04:30] *** CarlosC has joined #smartos[17:04:52] *** CarlosC has quit IRC[17:05:10] *** CarlosC has joined #smartos[17:07:12] *** nikolam has joined #smartos[17:11:52] *** wdent has quit IRC[17:20:00] *** joshie has quit IRC[17:23:19] *** tonyarkles has quit IRC[17:25:04] *** kamilr has quit IRC[17:25:28] *** ryancnelson has joined #smartos[17:32:59] *** wdent has joined #smartos[17:35:51] *** dubban has quit IRC[17:48:56] *** dap has joined #smartos[17:51:59] <alcir> ma[17:52:09] <alcir> to configure in the global zone[17:52:17] <alcir> another network interface?[17:52:31] <alcir> vi /usbkey/config[17:54:52] *** tonyarkles has joined #smartos[17:57:38] *** rc101 has quit IRC[18:06:12] *** deraps has quit IRC[18:14:00] *** notmatt has joined #smartos[18:21:42] *** KermitTheFragger has quit IRC[18:26:05] *** montyz0 has joined #smartos[18:27:46] *** dubban has joined #smartos[18:28:46] *** alcir has quit IRC[18:30:50] *** e^ipi_laptop has joined #smartos[18:31:11] *** jelmd has quit IRC[18:46:04] *** blahfop has joined #smartos[18:53:00] *** blahfop has quit IRC[18:54:35] *** dubban has quit IRC[18:55:47] *** dubban has joined #smartos[19:01:24] *** joshie has joined #smartos[19:02:33] *** ryancnelson has quit IRC[19:05:10] *** riceo_ has quit IRC[19:08:51] *** deraps has joined #smartos[19:22:29] *** notmatt has quit IRC[19:35:16] *** jim80net has quit IRC[19:45:20] *** mamash has joined #smartos[19:45:32] *** bens1 has joined #smartos[19:49:29] *** fsteinelX has joined #smartos[19:50:33] *** fsteinelX is now known as fsteinel[19:52:51] *** jesser has joined #smartos[19:56:26] *** CarlosC has quit IRC[20:04:13] *** kfr- has quit IRC[20:15:00] *** mamash has left #smartos[20:21:28] *** tonyarkles has quit IRC[20:25:46] *** e^ipi_laptop has quit IRC[20:29:27] *** Cpt-Oblivious has joined #smartos[20:29:40] <Cpt-Oblivious> Is it possible to use RAM as write cache in ZFS?[20:30:09] <Cpt-Oblivious> I know that when you have a power outage, that data will be lost. But if you place your server behind a UPS, this shouldn't really be a problem.[20:30:56] <Cpt-Oblivious> As far as I've been able to find out, you can only use ZIL as write cache and this can only be placed on a seperate disk, like an SSD, or a ZeusRam cache. That zeusram cache is ridiculously expensive though. So can ZFS also natively use RAM for write caching?[20:36:41] <wesolows> yes, effectively[20:37:13] <wesolows> zfs set sync=disabled[20:37:38] <wesolows> expect rampant data loss and corruption. have fun running your system "Linux-style"![20:38:50] <wesolows> and, bluntly, I'm not sure where the idea came from that only ZeusRAM can be used. While ZeusRAM is an excellent product and one I highly recommend where appropriate, it is as you note very expensive; there are cheaper (and slower) alternatives that nevertheless are far better than nothing.[20:39:06] <Cpt-Oblivious> I see[20:39:42] <Cpt-Oblivious> the only way RAM can be used for write caching is if you keep a ZIL log on ssd's then?[20:40:05] <Cpt-Oblivious> So that if a power outage occurs, in ZIL you'll be able to see how much from RAM had been written to disks already. So you don't get corruption, at most some data loss?[20:40:08] *** xinkeT has joined #smartos[20:40:35] <wesolows> the ZIL always exists, whether you have a dedicated log device or not.[20:40:41] <wesolows> if you don't, it's in the pool itself.[20:41:03] <Cpt-Oblivious> Hmmm[20:41:05] <wesolows> normally, when sync=default, transaction state is written into the ZIL at the time of the O_DSYNC write or fsync(2) call.[20:41:29] <wesolows> if you have no dedicated log device, that causes writes into the pool, comprised of spinning disks, which is slow -- ms to 10s of ms.[20:41:45] <wesolows> if you do have a dedicated log device, it's used instead.[20:42:13] *** Teknix has quit IRC[20:42:22] <Cpt-Oblivious> I see.[20:42:23] <wesolows> Later, once the relevant data has made it to the pool as a whole, the space used in the zil is freed up and reused for later records. ZIL contents are read only when the pool is being imported.[20:42:52] <wesolows> So there's really no such thing as using RAM for the ZIL, other than as an implementation detail in e.g. ZeusRAM, which happens to be an NVRAM device.[20:43:13] <Cpt-Oblivious> Okay.[20:43:20] *** jesser has quit IRC[20:43:30] <wesolows> sync=disabled simply holds that transaction state in memory as if O_DSYNC were clear, or fsync(2) were not called, then later writes it out to the pool as it normally would have.[20:43:45] <Cpt-Oblivious> So you can't use your normal system RAM as write cache, if you could make that non volatile with a UPS basically?[20:43:48] <wesolows> obviously, that's a gross violation of standard behaviour and one that will cause serious data loss.[20:44:15] *** bens1 has quit IRC[20:44:19] <wesolows> well, you sort of are, it's just not functioning as an intent log -- it's just got the state that it would have had anyway, waiting to be written out to the pool.[20:44:30] <Cpt-Oblivious> Ah I see[20:44:39] <wesolows> in other words, DRAM holds your data always.[20:45:08] <wesolows> when you (a) use sync=default, and (b) use O_DSYNC or fsync(2), your data will also immediately be written to something stable (i.e., which survives power loss)[20:45:25] <wesolows> when you instead use sync=none, that doesn't happen, even if you use O_DSYNC or fsync(2).[20:45:34] <Cpt-Oblivious> ah I see[20:45:45] <wesolows> that makes things super fast, but also means you're no longer getting the semantics you're asking for when you use the flag or the syscall[20:46:13] *** jim80net has joined #smartos[20:46:28] <wesolows> if you really want things to go faster, just change your applications not to do synchronous I/O. then at least each application can take responsibility for its own data loss.[20:46:39] <Cpt-Oblivious> if i remember correctly, fsynch is a call a program or application can do to the file system to ensure it's data has arrived on non volatile storage right? So, if you have synch=default, it will return an okay, when you issue a fsynch call. And the application will know it's safe. But if synch=none, it will issue an okay immediately regardless of whether it's on non volatile storage already.[20:46:57] <wesolows> yep[20:48:01] <Cpt-Oblivious> Advantages are, that if you use a PSU, you will have very fast write performance. and you'll be protected against power loss. But if you have hardware failure or the OS kernal that hangs, you'll still get massive data loss / corruption, right?[20:48:08] <Cpt-Oblivious> if you use a UPS*[20:49:32] <wesolows> right, a UPS only protects against a specific form of hardware failure[20:49:57] <wesolows> it won't save you from a mainboard fault, a kernel panic, or a bad CPU[20:50:13] <wesolows> also, UPSs are relatively expensive for what they do.[20:50:35] <Cpt-Oblivious> True, but you only need one that keeps it alive for 15 minutes and gives a gracefull shutdown command.[20:51:03] <wesolows> for the same price as a decent midrange UPS good for 20 minutes of uptime, you can buy an SSD that will provide you with persistent storage and good performance as a slog.[20:51:16] <Cpt-Oblivious> True.[20:51:28] <Cpt-Oblivious> And i agree, using an SSD as zil is probably the best option for most people.[20:51:32] <wesolows> the SSD will last longer, require no service, and protect you against a much wider range of failures[20:51:45] <Cpt-Oblivious> I was just wondering theoretically if you could use RAM for write cache.[20:52:03] <wesolows> of course. it's just software.[20:52:24] <wesolows> I mean, if you really want to do specifically what you're saying, just add a ramdisk to zfs as a log device.[20:52:25] *** papertigers_ has joined #smartos[20:52:42] <Cpt-Oblivious> what kind of ramdisk?[20:52:54] <wesolows> it's inefficient, dangerous, and probably won't work well across reboots, but it'll prove that it's possible :-)[20:52:57] <Cpt-Oblivious> a seperate physical disk with UPS build in, like the ZeusRAM?[20:53:16] <wesolows> I just meant ramdisk(7d)[20:53:38] <Cpt-Oblivious> ah, so you boot the OS, configure a ramdisk with what ever software, then set up your ZFS file sytem, and appoint that ramdisk as ZIL[20:53:40] <wesolows> again, you shouldn't do this. it's strictly worse than any of the other things I've described.[20:53:48] <wesolows> as a slog, yes[20:53:53] <Cpt-Oblivious> but yea[20:53:57] <Cpt-Oblivious> that'd be horrible :p[20:54:06] <wesolows> but if you're interested in theory, well there's theory for you.[20:54:09] <Cpt-Oblivious> since that disk clears on a reboot regardless and will always cause corruption[20:54:49] <Cpt-Oblivious> while setting synch=disabled. Will at least cause normal write back on a normal reboot.[20:55:06] <wesolows> what do you mean by normal writeback?[20:55:10] <Cpt-Oblivious> Only that is vulnerable too is, kernal panic / hardware failure.[20:55:31] <wesolows> if the pool is exported, there shouldn't be any log records to replay.[20:55:39] <Cpt-Oblivious> I'm assuming that if you put synch=disabled. That if you reboot the OS, ZFS will first write everything it has in cache to disks before it reboots?[20:55:53] <wesolows> depends on how you reboot it[20:56:16] <wesolows> personally I always use uadmin 2 1, which I can assure you does not cause ZFS to do any such thing[20:56:37] <wesolows> if you unmount and export the pool, then yeah.[20:56:49] <Cpt-Oblivious> ah ok[20:56:55] <Cpt-Oblivious> Yea that's a bit of a hassle[20:57:00] <Cpt-Oblivious> so that's definitely not worth it.[20:57:18] <wesolows> just use an SSD that's durable across power loss and call it a day[20:57:26] <Cpt-Oblivious> yep[20:57:30] <Cpt-Oblivious> probably the best option.[20:57:41] <Cpt-Oblivious> was just wondering :P[20:58:16] *** AlainODea_ has joined #smartos[20:59:01] <Cpt-Oblivious> Thanks for explaining anyway / clearing it up![21:02:41] <wesolows> np, good luck.[21:10:47] *** jamesd__ has joined #smartos[21:13:56] *** nikolam_ has joined #smartos[21:14:18] *** jamesd_ has quit IRC[21:15:28] *** wolstena has joined #smartos[21:15:54] *** nikolam has quit IRC[21:22:16] *** kfr- has joined #smartos[21:29:58] *** socketwi has joined #smartos[21:30:43] *** wdent_ has joined #smartos[21:31:21] *** e^ipi_ has joined #smartos[21:31:48] *** elijah-m_ has quit IRC[21:32:05] *** jim80net has quit IRC[21:32:23] *** elijah-mbp has joined #smartos[21:33:21] *** killfill_ has joined #smartos[21:33:25] *** elijah-mbp has quit IRC[21:33:49] *** nikolam_ is now known as nikolam[21:33:58] *** elijah-mbp has joined #smartos[21:34:04] *** dap1 has joined #smartos[21:34:43] *** jim80net has joined #smartos[21:35:30] *** killfill has quit IRC[21:35:32] *** dap has quit IRC[21:35:32] *** wdent has quit IRC[21:35:33] *** socketwiz has quit IRC[21:35:34] *** Mareo has quit IRC[21:35:35] *** e^ipi has quit IRC[21:35:37] *** e^ipi_ is now known as e^ipi[21:36:08] *** Mareo has joined #smartos[21:41:25] *** enmand has quit IRC[21:49:26] *** montyz0 has quit IRC[21:49:27] *** rancor has quit IRC[21:49:30] *** Azbruh has quit IRC[21:57:01] *** Mouaku has joined #smartos[21:57:12] *** rancor has joined #smartos[21:57:36] *** neophenix has quit IRC[21:58:03] *** rcamera has joined #smartos[21:58:48] *** ryancnelson has joined #smartos[22:00:52] <Mouaku> i am running smartos through virtualbox on my machine which has a i5 processor with vt-x and ept, but when booting i get the message kvm: no hardware support. does anyone know why this could be?[22:01:26] <Cpt-Oblivious> because it's a vm inside a vm.[22:01:40] <Cpt-Oblivious> if you run smartOS natively on your i5 proccessor, you won't get that.[22:01:49] <ryancnelson> yeah. your virtual cpu doesn't have vt-x or ept[22:01:57] <rmustacc> Mouaku: Because virtualbox doesn' emulate vt-x and ept.[22:02:32] *** AlainODea_ has quit IRC[22:03:38] <Mouaku> oh so the vt-x option inside virtualbox doesn't allow my guest os to use the tech...[22:05:00] <rmustacc> No.[22:05:04] <rmustacc> It allows virtual box to use the tech.[22:05:46] <Mouaku> is there no hosted virtualization which can pass it onto a guest?[22:05:49] <ryancnelson> i don't think there is such a thing as "vt-x option inside virtualbox" … that option is "let virtualbox use vt-x"[22:06:02] <ryancnelson> vmware fusion 5 on macs can do it.[22:06:56] <Mouaku> is that available on linux too?[22:07:12] <ryancnelson> no.[22:07:19] <Mouaku> :([22:07:52] <Mouaku> ah well guess ill have to put up with the slow speeds on my kvms while i mess around[22:08:02] <Mouaku> thanks for your help[22:12:40] *** kschiess has quit IRC[22:14:14] <papertigers_> Mouaku: kvm on linux allows nested VM's it may expose those options, not sure though[22:14:43] <papertigers_> The feature is not really used often by anyone so I don't know how it works[22:15:57] <ryancnelson> it's not used by anyone because it's terrible. don't expect not "slow speeds" if you're nesting virt like that… you're just lying to yourself, which has it's place sometimes, but you can't make something be something it's not.[22:16:08] *** socketwi has quit IRC[22:16:23] <ryancnelson> you can emulate a sparc processor with qemu, but it won't ever be a sparc processor.[22:16:29] <ryancnelson> for example[22:17:11] <ryancnelson> why not just boot into smartos for awhile, on real hardware? i swap disks to do just that, frequently.[22:20:09] <papertigers_> SmartOS is the way to go anyways. os virt and kvm dir[22:20:17] <papertigers_> virt work perfectly[22:20:18] *** mamash has joined #smartos[22:22:10] <Mouaku> i dont want to boot into smartos as i only have one computer and i want to run browsers etc[22:22:22] *** wdent_ has quit IRC[22:25:16] *** marsell has quit IRC[22:26:16] *** bens1 has joined #smartos[22:28:35] *** marsell has joined #smartos[22:29:04] *** iyp has joined #smartos[22:29:59] *** rcamera has quit IRC[22:30:12] <iyp> Trying to install freetds with odbc support. The package registered in pkgin is compiled with the --disable-odbc option. Any advice on getting a package not compiled with that option?[22:30:58] *** rcamera has joined #smartos[22:35:40] *** bens1 has quit IRC[22:37:43] *** marsell has quit IRC[22:43:35] <ryancnelson> if you can see the compile switches (see "pkg_info -B <pkgname> " ) , why not just change the one thing you don't like, and re-compile?[22:44:05] <ryancnelson> pkgsrc is just a collection of stuff someone got to compile, and shared the "how"[22:47:27] *** dubban has quit IRC[22:47:49] *** dubban has joined #smartos[22:50:29] *** fsteinel has quit IRC[22:50:34] *** montyz0 has joined #smartos[22:50:34] *** Azbruh has joined #smartos[22:54:33] *** jesser has joined #smartos[23:14:02] *** Tekni has quit IRC[23:14:12] *** tonyarkles has joined #smartos[23:17:58] *** Teknix has joined #smartos[23:21:08] *** iyp has quit IRC[23:28:25] *** tonyarkles has quit IRC[23:28:53] *** tonyarkles has joined #smartos[23:30:00] *** tonyarkles has quit IRC[23:35:09] *** mamash has left #smartos[23:39:53] *** montyz0 has quit IRC[23:46:52] *** xer0x has joined #smartos[23:58:24] *** nikolam has quit IRC