July 8, 2011 
< | 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 | >
July 8, 2011 [00:04:46] <DanaG> Now, is there an equivalent of Aptitude (the ncurses package manager)?[00:05:01] *** skeeziks has joined #openindiana[00:08:49] <alanc> no, the pkg manager is provided in CLI or GUI forms, no curses version[00:10:34] <DanaG> Bummer.[00:14:49] <OniAtWork2> although that wouldn't be a bad project for someone who wanted to do to do[00:15:17] <DanaG> Is it possible to create a zfs pool with an old version of zfs, intentionally?[00:15:33] <miine> DanaG: yep. man zpool :-)[00:16:12] <DanaG> Cool.[00:16:31] <miine> DanaG: if it makes sense to do so is a different thing...[00:17:01] <OniAtWork2> it does. It can be useful for moving data between versions of solaris[00:17:49] <OniAtWork2> I actually encountered a situation that It would have been useful last week (well, what would be REALLY useful is to just do some updates)[00:22:06] *** freedomrun has joined #openindiana[00:22:08] <DanaG> That "time slider" sounds very awesome.[00:23:10] <miine> DanaG: is it working?[00:24:45] <miine> DanaG: you're using OI for desktop or as server?[00:24:58] <DanaG> Here's the odd thing: what seems to have made the boot bearable is the act of disabling the graphics console kernel parameter.[00:25:10] <DanaG> I'll be using it for a server -- same use case as FreeNAS, but less crippled.[00:25:38] <DanaG> But having local (or kvm-over-ip) console will be helpful.[00:25:43] <miine> DanaG: then you'll should have a big /var/log/X11* log ...[00:26:31] <DanaG> Why would it make a big log?[00:26:38] *** gea__ has quit IRC[00:26:38] <miine> DanaG: console? you can setup a serial connection within grub, or you can get the desktop via X11...[00:27:19] *** davenz has joined #openindiana[00:27:23] <miine> DanaG: sorry. thought that you disable gdm (aka X11 login). what hardware do you use?[00:28:16] <DanaG> I did initially disable it, but now I redid the install.[00:28:21] *** descipher has quit IRC[00:28:23] <DanaG> The install to USB stick never succeeded.[00:28:34] <DanaG> What did work: install to vbox hard drive, then use linux to dd to flash drive.[00:29:01] <DanaG> Right now, it's booted on a supermicro h8smi, but the target system is the HP microserver.[00:29:04] <miine> DanaG: I would say that is a wrong concept somehow.[00:29:08] *** descipher has joined #openindiana[00:29:14] *** CoilDomain has quit IRC[00:29:24] *** CoilDomain has joined #openindiana[00:29:51] <miine> DanaG: to have something really working on usb take a look how the live-cd works.[00:30:08] <miine> DanaG: you can construct yourself one and extend it.[00:30:14] *** axisys has quit IRC[00:30:17] <DanaG> miine: I want the USB to be a permanent install, not an on-the-fly.[00:30:22] <miine> DanaG: google for distribution constructor[00:30:30] *** CoilDomain has quit IRC[00:30:31] <DanaG> I'd rather not have to rebuild every time I want to change something.[00:31:13] <miine> DanaG: you're doing a writeable system on an usb stick. doesn't sound very clever as no linux distribution does so too...[00:31:33] <DanaG> I've installed Ubuntu to USB before; it works fine.[00:31:55] <miine> DanaG: I would not rely on it in that way.[00:33:02] <miine> DanaG: and do not forget: even on live-cd /etc/ ... is writeable as it is held in ram. nothing hinders you changing the scripts and mounting a zfs dataset instead of the /tmp ram disk[00:33:30] <DanaG> But how about all the stuff not in /etc or /tmp?[00:33:43] <dkeav> put it on your zfs dataset[00:34:04] <DanaG> Ah, so I'd have a portion of the system lie on the hard drives.[00:34:14] <DanaG> How much would have to be on the flash drive?[00:34:35] <DanaG> And it sounds like a complicated way of doing things...[00:34:49] <miine> DanaG: you have to decide: either a working oi and one disk less if you want frequent system changes - or something which might work but might also be a shoot into your foot...[00:34:53] <tsoome> if half of your system is on hdd anyhow, whats the point of flash?:P[00:36:17] <dkeav> warm fuzzy feeling?[00:36:41] <tsoome> later its cold and... :D[00:36:50] <miine> DanaG: if I would be you (and you sound like an solaris/oi newbee) - just get it running firsthand and use a boot-hd and the remaining (3?) disks for your data...[00:37:20] <DanaG> https://www.illumos.org/issues/829[00:37:24] <miine> DanaG: later, if you really know whats going on in oi you might be able to get something working from usb[00:37:50] <DanaG> I'm very familiar with Linux, but totally new to Solaris, yes.[00:38:11] <tsoome> tbh, grub should be able to read compressed data instead:P[00:38:13] <DanaG> It seems to be now working (though not all that quick on disk operations) from USB.[00:38:16] <miine> DanaG: it will be even easier then because if you don't need to change the installed software on the running system it's easy as cake to prepare an usb stick with all desired packages on it...[00:39:34] <DanaG> Say, does this kernel support TRIM?[00:39:54] <DanaG> One option I'm considering is using a real SSD for boot drive, and a piece of it for cache drive.[00:40:27] <miine> DanaG: you can install/boot from whatever you want. you "just" have to edit the correct scripts. you'll have to look in which order the services are started and inside their scripts. thats a lot of work...[00:40:44] <tsoome> trim support in kernel is only an half of the game, its useless if filesystem does not call it:P[00:42:43] <miine> DanaG: if you really need to boot from usb (for whatever reason) - take my advise and look at the distribution constructor. even if you've to wait 2 hours for a build, the result makes it worth waiting...[00:45:13] <dkeav> do not try to run a cache drive from a boot drive[00:45:15] <dkeav> just plain dont[00:45:17] <dkeav> that is all[00:48:19] <DanaG> ah. Too much wear, for one.[00:54:29] <miine> DanaG: another thing: the live medium has NO zpool !![00:57:03] <miine> DanaG: BUT - if you dd the live-cd to the usb stick and mount it you can write to some places (like root users home content)...[01:00:09] *** skeeziks has quit IRC[01:01:16] <miine> DanaG: btw: if I remember correctly BeneliX is an OpenSolaris distribution which specialized on running from live-mediums (like usb sticks etc.).[01:05:19] *** gervystar has joined #openindiana[01:09:24] <DanaG> hmm, so maybe that's a viable way to do it.[01:09:39] <DanaG> How about installing packages (such as Deluge)?[01:10:17] <miine> DanaG: tell the distribution constructor to include them on the live-medium...[01:10:42] <miine> DanaG: you "just" have to find a way to get rid of the "language and keyboard" questions on startup.[01:10:42] <DanaG> Hmm, most specifically, the things I'd want are Deluge and some DAAP sharing server.[01:11:10] <DanaG> And Deluge isn't packaged for OI.[01:11:52] <miine> DanaG: you can modify directly, but it is a good idea to know how ips packages are constructed...[01:12:34] <miine> DanaG: hope you're not getting frustrated using oi...[01:14:07] <DanaG> Well, now that it's booting sanely, I'm less frustrated.[01:14:33] <DanaG> Before, it was taking a godawful long time, and the console install didn't even install properly -- it didn't create my user![01:14:59] <miine> DanaG: another idea: install those thing in zones on your hd. export them (on your "constructing" oi installation) and when booting from usb import and boot the zones.[01:15:39] <miine> DanaG: not creating the admin account? doesn't sound good. I would not trust that installation...[01:16:39] <miine> DanaG: do you have deluge and sharing stuff running on oi already?[01:17:29] <DanaG> miine: that install, I've overwritten with the clone from vbox.[01:17:57] <DanaG> And the clone was a GUI install.[01:18:47] <miine> DanaG: you waste your own time.[01:19:34] <miine> DanaG: want to hear my advice?[01:19:35] *** radsy has joined #openindiana[01:20:45] <DanaG> The text install is what I couldn't log into.[01:20:54] <DanaG> user "dana" didn't exist.[01:21:22] <DanaG> Root password wasn't set. Wasn't "root" / "OpenIndiana", either. Nor did "jack" / "jack" work.[01:21:31] <miine> DanaG: ok, if you want to hear it or not - I'll tell you:[01:21:46] <miine> DanaG: first - make an oi GUI install in vbox[01:22:08] <DanaG> That's done already, though I'll need to make the drive bigger.[01:22:18] <miine> DanaG: second: get the software compiled and running (this might keep you up for some time)[01:22:31] <DanaG> "the software" being deluge?[01:23:06] <miine> DanaG: yep. as there seems no oi release for it you have to learn the tricks to get it compiled and linked...[01:24:05] <miine> DanaG: then you can still decide if you want to have the trouble of having a usb boot OR just use the server[01:24:27] <miine> which means installing and booting from an hd[01:24:39] <tomww> there is SFEdeluge.spec, tough migh tneed a version updated and some love for the spec file[01:25:04] <miine> the hp microserver should have 4 bays (?), so you still should have enough room for the data.[01:25:39] <miine> if you can not - under any circumstances - "loose" a drive, THEN and ONLY THEN get yourself familar with usb booting[01:26:14] <miine> the BEST and ONLY reliable way is to use THE SAME WAY as the OI live medium does[01:26:34] <miine> and those are build with the distribution constructor.[01:27:10] <miine> it is possible to build your own live medium with the software you want "preinstalled" and configured.[01:27:36] <DanaG> ah, so you do the constructor from under the GUI install?[01:27:45] <miine> but - as I mentioned before - you have to edit the scripts which asks you for the language and keyboard layout[01:29:11] <miine> DanaG: distro constructor doesn't mind if text or gui install. but it is easier for people not so familiar with oi/solaris to use the GUI as you have also some other usefull tools on it...[01:30:21] <miine> DanaG: and it simply doesn't matter for you later if you use the gui or text live medium. the difference would be using a 1 GB or 2 GB flash drive...[01:31:45] <miine> DanaG: but take my words (I know what I'm speaking about as I've spend much time on that stuff): FIRST get the things running on a normal HD.[01:32:22] <DanaG> Just have to find a spare drive that's healthy and low-power.[01:33:03] <miine> DanaG: there might be some "surprises" you didn't expect, and knowing that the "basics" are right and working is the only way to isolate other failures...[01:34:06] <miine> DanaG: just a quick question: if I remember correctly the microserver has a broadcom nic?[01:34:25] <DanaG> Yeah, but I just purchased an Intel since it's cheaper today.[01:34:28] <DanaG> Should get it tomorrow.[01:34:36] <DanaG> I mean, cheap today.[01:34:49] <miine> DanaG: if so - you MUST update the bge driver as will do a LOT of interrupts and slow everything down...[01:34:58] <miine> DanaG: Intel is a good thing :-)[01:35:52] <lennard> myricom is even better :)[01:36:31] <DanaG> Yeah. I once copied from Core2Duo + Intel + Vertex-2 to Pentium M + Broadcom + OCZ-Vertex, and got 80MB/second. 100% CPU usage on the Pentium M.[01:38:41] *** skeeziks has joined #openindiana[01:40:45] *** freedomrun has quit IRC[01:43:34] <DanaG> Hmm, so assuming I go with regular hard drive, how much does peformance matter? is an old 5400rpm drive enough?[01:44:04] <miine> DanaG: doesn't matter :-) as the most time things will be in RAM anyway...[01:44:53] <miine> DanaG: I'm using here an Atom 330 with an 2.5 HD drive for boot and 4x5400 RPMs RAID-Z1 for data[01:45:35] <DanaG> Hmm, so what is the problem with USB + flash drive: is it the "flash drive" part, or the "USB" part?[01:45:47] <miine> DanaG: its not a speed wonder (as 2 of the RAID-Z1 drives are connected via PCI controller :-( ), but its fast enough for moderate file sharing, email, voip...[01:46:23] <miine> DanaG: the flash part[01:46:31] *** magyar has quit IRC[01:47:30] <miine> DanaG: so your Microserver should be fine for up to 5 people using it at the same time...[01:47:33] *** magyar has joined #openindiana[01:49:23] <miine> DanaG: or maybe even ten. just get it running first. think more about how you want to configure your zpool with the remaining disks as shuffling data around later might be painfull (if there is no big enough drive to copy to...)[01:51:05] <DanaG> Right now, I have a single, 2TB NTFS drive, 3/4 partitioned, and 3/4 of that used. I also have a second 2TB drive, and a third one on the way.[01:51:13] <DanaG> So I have no specific plans.[01:51:24] *** magyar has quit IRC[01:52:01] <miine> DanaG: so either do a RAID-Z1 or RAID-Z2 (4 or 2 TB usable). depends on how many drives you want to be able to loose at the same time vs. space needed...[01:52:38] <DanaG> I'm thinking I'll want some sort of offsite piece.[01:52:46] <DanaG> SO I could mirror, and add offsite.[01:53:53] *** kfritz has joined #openindiana[01:55:34] *** kfritz has quit IRC[01:57:29] <miine> DanaG: just wondering - is the DVD-ROM within the microserver connected via SATA?[01:57:51] <DanaG> Right now it is. Hey, that's a good idea: move that to USB.[01:58:15] <miine> DanaG: remove it at all and put there in a HD :-)[01:58:25] <DanaG> Nah, a 2.5 can fit in the gap below that![01:58:51] <miine> DanaG: did that with an Dell T110 to have 5 drives...[01:59:27] <miine> DanaG: I would tend to have some airflow for spindles...[02:06:29] <miine> is there an easy way to find out the process(es) accessing a specific disk?[02:07:36] <DanaG> The other thing with hard drives is that I have very few non-noisy spares.[02:07:55] <DanaG> Lots of 40GB Hitachi Travelstar PATA drives, and not much else.[02:09:27] <miine> DanaG: if done right you won't access the boot hd that much... .[02:09:56] <miine> DanaG: so I would care only about noise if I would buy a new disk[02:11:26] *** McBofh has joined #openindiana[02:12:51] <miine> DanaG: on my Atom 330's rpool there is just 14.1G used of 60G, and from those there are 4 GB oi repository...[02:15:32] *** skeeziks has quit IRC[02:19:53] *** skeeziks has joined #openindiana[02:19:54] <DanaG> ah, gotta' go for now.[02:20:00] <DanaG> But thanks for the tips.[02:20:05] *** DanaG has quit IRC[02:20:10] *** skeeziks has quit IRC[02:30:49] *** tomww has quit IRC[02:31:01] *** tomww has joined #openindiana[02:31:01] *** ChanServ sets mode: +o tomww[02:31:02] *** magyar has joined #openindiana[02:31:02] *** magyar has joined #openindiana[02:34:46] *** InTheWings has quit IRC[02:38:17] *** movement has quit IRC[02:40:51] *** axisys has joined #openindiana[02:45:59] *** DrLou has quit IRC[02:46:13] *** Naresh` has quit IRC[02:47:50] *** Naresh` has joined #openindiana[02:50:06] *** movement has joined #openindiana[03:11:01] *** master_of_master has quit IRC[03:17:22] *** ben_sysmgr has joined #openindiana[03:21:03] *** master_of_master has joined #openindiana[03:22:05] *** axisys has quit IRC[03:24:59] *** ben_sysmgr has left #openindiana[03:33:08] *** redgone has joined #openindiana[03:34:10] *** radsy has quit IRC[03:34:34] *** axisys has joined #openindiana[03:37:13] *** gervystar has quit IRC[03:38:54] *** Gnu_Raiz has joined #openindiana[03:56:06] *** DanaG has joined #openindiana[03:57:03] *** axisys has quit IRC[04:01:53] *** axisys has joined #openindiana[04:06:54] *** axisys has quit IRC[04:07:16] *** dijenerate has joined #openindiana[04:07:49] *** axisys has joined #openindiana[04:12:04] *** axisys has quit IRC[04:15:21] *** Oriona has quit IRC[04:15:46] <DanaG> Say, given two drives I want to pair, which is better to use: raidz, or mirroring?[04:15:53] <DanaG> I'd assume mirroring, because it's simpler.[04:15:57] *** axisys has joined #openindiana[04:16:22] <DanaG> I was going to try to nondestructively add one new drive to an existing non-zfs drive... but then I realized it's easier to just buy a third drive -- and I'd be wanting it anyway.[04:17:39] <OniAtWork2> I would mirror[04:18:47] *** POloser has joined #openindiana[04:22:01] *** mnaser has quit IRC[04:26:22] *** xmikus01 has quit IRC[04:29:35] *** axisys has quit IRC[04:30:14] <sivanov> interesting: i get 300Mbit/s iperf to freebsd server while the network interface is vacant and 550Mbit/sec while interface is busy, serving ISCSI to another host while it runs iozone on the iscsi LUN[04:31:06] *** axisys has joined #openindiana[04:33:37] <DanaG> Say, would an slc SD card be a reasonable boot drive?[04:33:46] <DanaG> Assuming I used a single-slot sd-only reader.[04:34:37] <sivanov> i'd avoid using usb to boot a server[04:34:46] <DanaG> One of these cards: http://www.newegg.com/Product/Product.aspx?Item=N82E16820157058[04:35:02] *** axisys has quit IRC[04:35:19] <sivanov> i'd rather buy cheap 30Gb sata2 ssd[04:37:38] <DanaG> Hmm, wait... I just remembered, I have a 160GB 5400RPM notebook drive that's not horribly noisy.[04:39:37] <DanaG> I'll probably stick with the hdd for now, until I find something good and inexpensive.[04:41:10] <DanaG> All the SATA laptop drives I have, are in use somewhere.[04:41:16] <DanaG> No spares anywhere.[04:44:25] <DanaG> TOmorrow I'll move the dvd drive to usb, and the PATA hard drive onto a SATA adapter.[04:44:39] <DanaG> But anyway, what's wrong with an SLC SD on USB?[04:45:07] <DanaG> I get that MLC is bad due to wear limits... and my Patriot XPorter XT Boost is slow as molasses on OpenIndiana (yet is fine on Ubuntu, by the way).[04:50:55] <DanaG> Say, OpenIndiana seems to be missing a "radeon" kernel module -- not that that matters for a NAS.[04:55:55] <OniAtWork2> Last I new there wasn't an accellerated one avalible.[04:56:04] <OniAtWork2> that was ages ago though[04:57:09] <DanaG> hmm, does it have a driver for "ASPEED" chips?[04:57:43] <DanaG> ASPEED makes IPMI chips, like the one on the Microserver remote access card.[04:57:51] <taemun> tyan lom chip[04:58:26] <DanaG> I wish AMD would make an IPMI chip that had an ES3000 (or its successor) in it.[04:58:37] <DanaG> So at least then we'd get KMS on Linux. And 2D that didn't suck.[04:59:31] *** ilc has quit IRC[04:59:53] <taemun> ( DanaG ) But anyway, what's wrong with an SLC SD on USB? << the usb to sd chip has a high probability of being crap[05:12:13] *** kart_ has joined #openindiana[05:15:59] <DanaG> bah. I do wonder why OpenIndiana is so much slower on the same flash drive, versus Linux.[05:16:58] *** axisys has joined #openindiana[05:21:40] <alanc> there's Xorg drivers for the ASPEED AST2000/2100/2200 at least, since Sun shipped those as the ILOM in a bunch of servers - I assume there's drivers for the other components as well[05:23:14] <DanaG> Say, how do I make sure OpenIndiana is checking/reporting ECC?[05:25:30] <alanc> didn't we go through this a couple days ago?[05:26:18] <DanaG> fmdump shows just that it's empty.[05:26:41] <alanc> then you've not had any problems to report[05:26:46] <DanaG> The log, rather.[05:27:40] <alanc> it only reports once errors cross certain thresholds (different levels for different types of devices)[05:27:42] <DanaG> But how do I know it's even checking? I know in Linux, I get notice of amd64_edac loading. Oh, right... modinfo.[05:27:55] <richlowe> won't tell you anything specific.[05:28:07] <richlowe> if us telling you it's checking won't help, I don't know of better answers.[05:28:20] <alanc> neither do I[05:29:08] <richlowe> fmadm config[05:31:16] <DanaG> cpumem-retire disk-transport eft ext-event-transport fabric-xlate fmd-self-diagnosis io-retire sensor-transport ses-log-transport software-diagnosis software-response sysevent-transport syslog-msgs zfs-diagnosis zfs-retire[05:35:58] <sivanov> hmm, apparently, the bge driver does not like when i bind it to CPU 2-3[05:36:11] <sivanov> low performance via iperf[05:37:29] <sivanov> when i bind it to cpu 3 iperf gives me 185Mbit/s 100% of time[05:38:17] <DanaG> smbios says: ECC: 5 (single-bit ECC)[05:39:04] <sivanov> when i bind to cpu 0, iperf gives 939Mbit/s[05:40:46] <DanaG> memscrub_scans_done: 4e1 = loopz +0x4 <0xfffffffffbc80366>[05:41:07] <DanaG> https://developmentonsolaris.wordpress.com/2009/03/06/how-to-make-sure-memory-scrubber-is-running/[05:41:11] <DanaG> So that's it, I guess.[05:47:47] *** McBofh has quit IRC[05:50:32] *** McBofh has joined #openindiana[05:52:05] *** redgone has quit IRC[05:52:14] <DanaG> hmm, napp-it looks interesting.[05:56:24] <miine> DanaG: you said that you want to put 3 x 2TB into the server, two disks as mirror and one disk for "offsite" stuff.[05:56:44] <miine> DanaG: why don't you simply boot from the mirror?[05:57:19] <DanaG> Good point. I hadn't thought of that.[05:57:24] <DanaG> That'd probably give best performance.[05:57:42] <miine> DanaG: that WILL give the least headaches...[05:57:47] <DanaG> I plan to have two in the server always, and a third that I lug around.[05:57:57] <DanaG> Cool. Now, is there a guide for that already?[05:58:10] <DanaG> I probably don't need all that much space for the system partition, if you even have to do partitions.[05:58:16] <miine> DanaG: sure. just google for opensolaris and rpool mirror[05:58:18] <DanaG> I'm not familiar with the ZFS dataset concepts.[05:59:40] <miine> DanaG: you just have to make sure on installation that you DON'T choose "whole disk", otherwise you can't mirror later.[06:00:23] <miine> DanaG: but you WILL use the whole disk for the rpool which is then later mirrored...[06:01:41] <miine> DanaG: forget the last thing, will confuse you only. just make sure you won't check "whole disk" and use the installer default settings for partition size...[06:02:17] *** PMT has quit IRC[06:02:18] <DanaG> Do I do that with both disks in, but only one selected? And I'd leave my ntfs (to become offsite one)out.[06:02:50] <miine> DanaG: you can not install oi as mirror (as you can with solaris 11 ex). you install onto one and mirror that later. google ![06:03:00] <DanaG> Also, I may want to work with boot-from-separate-drive first, to make sure I'm okay with OpenSolaris. The other option was FreeNAS, which has an old pool version.[06:04:15] <miine> DanaG: if you fear to loose some data by accident that might be a good idea...[06:04:45] <miine> DanaG: but maybe you should learn about ZFS snapshots and Boot Environments.[06:05:15] <DanaG> The biggest things I'd want would be something like napp-it, and Deluge. That's about it.[06:05:32] <DanaG> So I'll try both now, before doing anything much with the new drives.[06:06:18] <miine> DanaG: and as your data is mirrored, you can remove a drive when powered off and put it aside when doing stuff that might be more experimental...[06:07:00] *** syoyo_ has joined #openindiana[06:07:39] <DanaG> The bigger thing, for now, is the pool version. Once I test napp-it and Deluge, I believe I can commit to OpenIndiana, and use current pool version.[06:16:20] <sivanov> btw, how i clone io install to another disk ?[06:16:21] <DanaG> Oh, and how do I make sure to keep partition boundaries aligned to 4K blocks?[06:16:38] <sivanov> different size[06:16:53] <miine> DanaG: you won't create partitions.[06:17:19] *** mit__ has quit IRC[06:18:21] <DanaG> http://lists.illumos.org/pipermail/developer/2011-January/001218.html[06:18:29] *** Naresh` has quit IRC[06:19:13] <miine> DanaG: you can try and try and try. or install napp-it and deluge and be happy...[06:20:20] <miine> DanaG: if you would have done more research on this 4k thing you would have found some infos about the speed difference (which doesn't matter so much)[06:20:49] <DanaG> I've done research before, and some cases found like 30% performance difference.[06:22:07] <miine> DanaG: and? your ethernet will be anyway the bottleneck... and I wouldn't also install some hacks which are not in oi by default...[06:23:28] <sivanov> DanaG, you need either zfs utility with ashift=12 patch, or check if that patch is in the io-148 already[06:23:47] <sivanov> *zpool utility[06:24:01] <DanaG> Anyway, I did go ahead and purchase the Intel card.[06:26:02] <sivanov> btw, i'm not sure if network is bottleneck: right now i'm trying to squeeze more MBps from iscsi target on x3650 ibm server. it has four gb ethernet links dedicated to iscsi, and i cannot get more that 55MB/s from it[06:26:59] <DanaG> I'll just have a pair of 2TB Samsung F4 drives.[06:27:03] <sivanov> the pool can do ~180MB/s on write and 350MB/s on read[06:27:06] <DanaG> Mirrored. And an Intel card.[06:27:20] <miine> sivanov: I think anyone would be quite happy with 55 MB/s from cheap disks and a HP microserver on a SINGLE gb link...[06:27:53] <sivanov> and the cpu is atom?[06:27:55] <miine> sivanov: so you have a problem with iscsi...[06:28:14] <DanaG> The highest I've gotten when testing gigabit: 80 megabytes per second. That's from Vertex 2 in Core 2 Duo laptop with Intel nic, to Vertex 1 in Pentium M laptop with Broadcom NIC.[06:28:32] <DanaG> sivanov: nope, an Athlon II Neo N36L. It's embedded, and yet does ECC.[06:28:33] <miine> sivanov: DanaG's hp microserver has an AMD...[06:28:49] <sivanov> nice[06:29:14] <sivanov> ddr3 ecc ram is VERY cheap on ebay[06:29:14] <miine> DanaG: you can't get much more then 80 MB on GB ethernet...[06:29:30] <DanaG> I weighed that against the SuperMicro atom board + case + 5-in-3 adapter... and the HP won.[06:29:40] <sivanov> you can buy a 2-4gb dimm for a dollar if your lucky[06:30:03] <DanaG> I bought one 4GB stick on Newegg... don't remember the price, but not that low.[06:30:12] <DanaG> That 80 was with 100% CPU usage on the Pentium M, by the way. 915GM chipset.[06:30:24] <sivanov> actually, ddr3 ecc is the cheapest ram on ebay[06:30:39] <DanaG> And that was Win7 to Win7, over CIFS.[06:30:46] <DanaG> Common Internet File System? Hah![06:30:51] <DanaG> I hope not.[06:31:02] <DanaG> No samba over the internet, please. =þ[06:31:59] <sivanov> is it possible to isolate all kernel stuff on cores 0-1 and all user stuff on cores 2-3 in OI?[06:32:10] <DanaG> What would running the OS from the data drives do, with regards to drives spinning up?[06:32:36] <miine> DanaG: what would you expect to do it?[06:33:38] <DanaG> My ideal behavior would be to keep the drives spun down as much as possible.[06:33:52] <DanaG> Then again, the PSU fan is like 3 times as noisy (not loud, but grating) as everything else.[06:33:56] <sivanov> its not possible with zfs[06:34:00] *** Hyphenex has left #openindiana[06:34:07] <miine> DanaG: so find out what might write/read to disk and minimize that.[06:34:07] <sivanov> or hardly possible without hacks[06:35:09] <miine> DanaG: when using consumer drives its a good idea not to have them 24/7 spinning and being a bit more to the drives specs...[06:35:36] <sivanov> i think not[06:35:58] <sivanov> the heads load/unload cycle will destroy the consumer hdd[06:36:57] <miine> sivanov: read their specs. consumer drives are prepared for that. server drives wouldn't like spinning up and down the whole day...[06:37:51] <miine> sivanov: of course it makes no sense to spin down after 10 secs or so. 15 min or something like that should be fine.[06:38:13] <sivanov> just remember how it was with some series of WD drives: they had the spin down time programmed too low to use in linux/unix/raids[06:38:18] <miine> just let the drive idle for 8 hours per day. should keep it healthy[06:38:50] <sivanov> after a month uptime the smart would count milliens load/unload cycles on them[06:39:01] <sivanov> and the drive would go bad[06:39:05] <DanaG> And their fix? Stop exposing that attribute.[06:41:12] <sivanov> and zfs will access the drives every munute of so[06:41:31] <madwizard> Coffee[06:42:09] <miine> sivanov: so you tell me that zfs won't let the drives spin down?[06:42:21] <miine> sivanov: or constantly wake them up?[06:42:29] <sivanov> wake up them[06:43:38] <miine> sivanov: ZFS does nothing of that. zfs does only access on somebody else's behalf. so find that culprit and the drives will sleep.[06:45:09] <DanaG> I hope there's something like 'iotop' -- iostat isn't it.[06:45:52] <sivanov> i found it when i was experimenting with disk spin down utility under freebsd. ufs formatted disk spins down nicely while one with zfs pool wakes up every munute[06:46:46] <sivanov> and read on internet that it has to do with how zfs handles zil and caches[06:46:53] <miine> sivanov: don't know what freebds implemented there. also don't know why people use zfs on freebsd at all when there was opensola ris ;-)[06:47:15] <sivanov> so zfs has flush caches every minute[06:47:39] <miine> sivanov: you think that zfs will flush an empty cache to disk every minute??[06:47:46] <sivanov> yes[06:48:11] <miine> sivanov: I don't think that the sun engineers are that stupid. otherwise we won't have zfs...[06:49:44] *** McBofh has quit IRC[06:52:30] *** McBofh has joined #openindiana[06:56:59] <DanaG> Anyone here used that napp-it?[07:04:13] <sivanov> but what to do with this iscsi: regardless if i use 1 link@fixed path or 4 links in round-robin i get same 55MB/sec read from it[07:05:10] <miine> sivanov: what do you get in virtual box as initiator?[07:05:29] <sivanov> this is pretty bad performance for gigabit & mtu 9000[07:05:35] <miine> sivanov: on same host (crossbow...) ??[07:05:45] <sivanov> initiator is ESX 4.1U1[07:06:13] <sivanov> run iozone in a linux VM[07:06:45] <miine> sivanov: test it on local host to see if the problem is the target or the initiator or the network between[07:07:02] *** jaquer has joined #openindiana[07:08:13] <sivanov> well, infotrend a16e-g2140 on same switch gives more than 100MB/s to the ESX host[07:08:23] <jaquer> Hi all. I'm moving from Debian Linux to OI for my fileserver. How do I prepare the disks to be used in the zpool? Should I wipe the partition table and use Solaris slices instead?[07:08:34] <miine> sivanov: what does oi in vb get as speed?[07:09:20] <miine> sivanov: we can assume that oi speaks with oi at good speed without tuning anything...[07:10:13] <sivanov> well, i can run OI on another server, but dont have 4 ethernet interfaces[07:10:47] <miine> sivanov: doesn't matter. google crossbow and vnics ...[07:11:25] *** McBofh has quit IRC[07:11:53] <miine> sivanov: but even without crossbow you can assign more then one nic to a vm in vbox...[07:12:36] <DanaG> hmm, that old PATA drive isn't much faster on SATA than on USB.[07:20:13] <DanaG> ImportError: No module named chardet[07:21:13] <DanaG> ah, downloaded it myself.[07:21:59] <DanaG> Sweet, deluge builds and installs and runs.[07:29:06] <kohju> Cool, http://www.youtube.com/watch?v=_s2NFNwNtEk[07:30:51] *** Botanic has quit IRC[07:32:49] *** descipher_ has joined #openindiana[07:33:54] *** descipher has quit IRC[07:34:00] *** descipher_ is now known as descipher[07:36:34] *** davenz has left #openindiana[07:41:34] *** jaquer has quit IRC[07:42:43] *** hsp has joined #openindiana[07:46:46] <sivanov> oh my, Athlon II Neo N36L is so slow...[07:47:17] <sickness> it's not supposed to be fast either...[07:48:31] *** freedomrun has joined #openindiana[07:58:17] <DanaG> It's supposed to be low power, but not crap like Atom.[07:58:44] *** McBofh has joined #openindiana[07:59:00] <sickness> heh[07:59:33] <DanaG> Atom would be less crap if it supported ecc.[07:59:54] <miine> atom isn't that bad - even without ecc.[08:00:17] <rj> yeah, it's not via nano![08:00:21] * rj crawwwwls[08:00:38] <miine> what missing there are decent motherboards (6x sata, IMPI, IP-KVM) which are affordable...[08:00:49] *** gea__ has joined #openindiana[08:03:04] *** McBofh has quit IRC[08:04:52] <miine> the hp microserver looks promising, but when thinking about pc hardware i want no special form factor mobos and powersupplies...[08:05:21] *** McBofh has joined #openindiana[08:05:25] <DanaG> That's a normal FlexATX.[08:05:26] <miine> would be like using a raid controller...[08:05:39] <DanaG> There's an IPMI card for that microserver.[08:05:50] <DanaG> FlexATX PSU, that is, not necessarily mobo.[08:07:07] <sivanov> Children see throughput for 4 rewriters = 34521.56 KB/sec[08:07:17] <sivanov> Children see throughput for 4 random writers = 23829.77 KB/sec[08:07:21] <sivanov> not very fast[08:07:40] <miine> DanaG: mobo should be a special form factor. so you either choose an extended warranty or it makes sense to drop it completly in case of failure...[08:07:51] <sivanov> its nfs over GigE[08:09:40] <DanaG> hmm, I honestly think I may go back to FreeNAS. I don't need the full power of OpenIndiana, and I'd like to have a web UI.[08:09:45] <DanaG> Thanks for all the help, though.[08:10:04] <miine> really astonishing is how less power an Intel i3 uses when idle... today I would buy one of those old Dell T110 instead of building my own atom box...[08:19:34] *** sponix has quit IRC[08:22:33] *** Botanic has joined #openindiana[08:28:27] *** xmikus01 has joined #openindiana[08:29:29] *** syoyo_ has quit IRC[08:32:15] *** gea__ has quit IRC[08:38:11] <sivanov> is it possible to mine bitcoins on OI?[08:38:53] <Wraithh> With cpu, sure. Dunno about opencl[08:45:47] *** magyar has quit IRC[08:46:26] *** alcir has joined #openindiana[08:51:31] *** edogawaconan has quit IRC[08:52:07] *** ry has joined #openindiana[08:52:17] *** sponix has joined #openindiana[08:52:25] *** dekar has quit IRC[08:53:49] *** magyar has joined #openindiana[08:53:50] *** magyar has joined #openindiana[08:56:02] *** kinesis has joined #openindiana[08:57:10] *** dekar has joined #openindiana[08:57:14] *** magyar has quit IRC[08:57:39] *** edogawaconan has joined #openindiana[08:58:54] *** |AbsyntH| has joined #openindiana[08:59:29] *** miine has quit IRC[09:02:27] *** magyar has joined #openindiana[09:06:18] *** Oriona has joined #openindiana[09:08:31] <gea_> @DanaG: I have wrote the Web-UI napp-it for OpenIndiana, no need to go back to FreeNAS :)[09:08:33] <gea_> see http://hardforum.com//showthread.php?t=1573272[09:09:15] <DanaG> Well, there were other things, like it being a pain to get things like Deluge and Transmission.[09:09:27] <DanaG> And the sucky performance of that USB drive, that works fine in Linux and BSD.[09:10:02] <DanaG> Now, if somebody got Deluge to compile, that'd be helpful.[09:10:09] <DanaG> Deluge, and some DAAP streaming thing.[09:10:38] <Oriona> gea_ Great work with Napp-it :D[09:11:00] <gea_> You may use OI as Storage machine and virtualize the rest on its best platform[09:12:51] <DanaG> Virtualize? I'd rather not have to admin two systems in one.[09:12:56] <DanaG> I'd rather just have it run natively.[09:13:26] <DanaG> I'll consider OpenIndiana again this weekend, when I have more time to try to compile stuff.[09:14:08] *** Micr0mega has joined #openindiana[09:14:21] <DanaG> The big two things are torrents (deluge, specifically), and daap (no specific preference on which server app).[09:16:33] *** sivanov has quit IRC[09:17:36] *** sivanov has joined #openindiana[09:18:18] <gea_> ESXi as a barebone virtualizer is quite easy to manage with a decent windows app and its really a nice thing to have multiple OS on one machine[09:18:34] <DanaG> Too bad HP doesn't allow PCI passthrough on the microserver.[09:18:41] <DanaG> The chip should support it, but their BIOS is sucky.[09:19:04] <DanaG> Somebody with a lot of influence (no idea who that would be) should encourage them to enable that -- and all the stuff the AHCI-mod BIOS allows.[09:19:27] <gea_> Chipset, CPU and bios must support, Microserver does not at all[09:19:31] <DanaG> (that is, hotplug ahci on all ports, instead of legacy-mode on two and non-hotplug on 4).[09:19:51] <DanaG> Athlon II Neo should support it, shouldn't it? At least, other Athlon II do.[09:20:34] <gea_> you need Intel vt-d or AMD IOMMU support[09:21:47] <DanaG> http://www.semiaccurate.com/forums/archive/index.php/t-2141.html[09:22:04] *** sivanov has quit IRC[09:22:30] <DanaG> http://forums.amd.com/forum/messageview.cfm?catid=383&threadid=134410&enterthread=y[09:23:27] *** sivanov has joined #openindiana[09:23:38] *** raichoo has joined #openindiana[09:24:39] <DanaG> Looks like it's a mess.[09:24:49] *** hajma has joined #openindiana[09:24:53] *** Botanic has quit IRC[09:24:58] <gea_> You can virtualize with ESXi on a HP microserver, i know of a lot of people who do[09:25:00] <gea_> but you cannot pass-through hardware (must use ESXi disks as datastore)[09:25:39] <DanaG> hmm, I'd love to get a clear answer from HP and AMD: is this a hardware limitation, or a BIOS limitation?[09:26:11] <gea_> as i know, its not bios but hardware[09:27:07] <DanaG> Anyway, if I can get torrent client and DAAP server to work, that's about all I'll need. No need to virtualize anything.[09:27:11] *** sivanov has quit IRC[09:27:29] *** xmikus01 has quit IRC[09:28:56] <DanaG> that, and I'd like to get that damn usb stick to work as decently as it does in other OSes.[09:32:11] <gea_> You cannot use USB and be happy on OI like you cannot do on Windows[09:32:13] <gea_> use a 2,5" sata disk or SSD (i know of limited sata slots on microserver, but it is as it is)[09:32:15] <gea_> you may use extra sata controller of course to have more slots but then you have bought a too small machine[09:32:17] <gea_> in this case its mostly better to replace disks with bigger ones[09:35:20] <gea_> (and use the old ones for backup via USB)[09:35:30] *** DanaG has quit IRC[09:42:01] *** freedomrun has quit IRC[09:51:54] *** McBofh has quit IRC[09:56:58] *** McBofh has joined #openindiana[10:12:58] *** McBofh has quit IRC[10:14:45] *** Whoopsie has joined #openindiana[10:14:45] *** ChanServ sets mode: +v Whoopsie[10:17:04] *** Naresh` has joined #openindiana[10:18:57] *** McBofh has joined #openindiana[10:29:17] *** sivanov has joined #openindiana[10:29:25] *** mikw has joined #openindiana[10:36:08] *** wonslung has quit IRC[10:36:47] *** wonslung has joined #openindiana[10:36:49] *** mikw has quit IRC[10:37:06] *** mikw has joined #openindiana[10:44:00] <kinesis> i have some ftp "server" on oi_148, based on Asus AT3IONT-I mb, and 2 seagate ST31000528AS hdds in mirror. write speed on pool is 10Mb/s max. reading near 100. what could be the cause?[10:45:26] <sickness> cache disabled on disks?[10:46:18] <kinesis> i think no. how can i found it?[10:50:15] <kinesis> format dont have cache option at all[10:50:38] *** gervystar has joined #openindiana[10:50:58] *** Naresh` is now known as Naresh[10:51:00] *** Naresh has joined #openindiana[10:51:28] <lblume> format -e[10:53:46] <kinesis> Write Cache is enabled[10:54:35] <lblume> Shouldn't matter anyway, zfs doesn't like write caches and flushes them all the time/[10:54:47] *** PMT has joined #openindiana[10:55:48] <kinesis> try to disable it? on both disks i think?[10:55:50] <lblume> Test the drive individually and from a different OS to help pinpoint the issue. I have a bunch of those disks with S10, they work well. I did upgrade their firmware, though.[10:56:16] <nettezzaumana> servus[10:57:47] <kinesis> but on other server all of 16 disks have cache enabled. and write speed is excellent.[10:57:56] *** vadim_comanescu has joined #openindiana[10:58:16] <kinesis> on other server i have hitachi hdds instead of seagate.[10:58:29] <vadim_comanescu> hi. is there any way i can make svccfg set multiple properties in one command on a service?[10:58:43] <lblume> kinesis: Like I said, I have those Seagate, they work, with write cache. It does not matter[10:59:10] <kinesis> i allready test disk individualy in windows, working fine.[10:59:31] <kinesis> may me it chipset on this MB?[11:00:05] <kinesis> sata configured in bios as AHCI[11:00:11] <lblume> not one of the broken Cougar Point?[11:01:10] <kinesis> will google for Cougar Point ) thanks![11:02:14] *** lkocman has joined #openindiana[11:02:27] <lblume> If it's an Intel i3/5/7 purchased early this year, chances are it's the buggy chipset, which has sata performance issues developing after a while, though I didn't keep mine long enough to find it.[11:02:52] <nettezzaumana> do you know what i love on these chans ? all the same everyday's buzz here .. geeks are geeking same things again and again .. oi is reduced to zfs-mainly platform - nothing more .. sometimes is someone astonished (or disguisted) by poor i/o performance by nothing real happens[11:03:11] *** befreax has joined #openindiana[11:03:17] <nettezzaumana> s/by/but/$[11:04:11] <lblume> Try #lulzcats for a change.[11:04:15] <nettezzaumana> difference between #solaris and #oi is that in first one ppl still perform task obvious to OS, here it's all about 2 things[11:04:19] *** ksx4system has quit IRC[11:04:31] <kinesis> AT3IONT-I have nvidia chipset with integrated intel Atom 330[11:05:18] <nettezzaumana> lblume: you robbed me up !! no #lulzcats exists[11:05:34] <lblume> Then create it! Time for you to step forward![11:06:03] <lblume> kinesis: Then that's not the issue. Anything in the logs?[11:07:07] <kinesis> lblume, nothing.[11:08:13] <lblume> bench it in IDE legacy mode, maybe[11:09:21] <kinesis> ok, will try it when i can stop this "server" )[11:10:26] <kinesis> but if i switch sata to ide legacy, how can i boot OI from this controller?[11:16:01] <lblume> you can't[11:16:06] *** vadim_comanescu has quit IRC[11:16:28] <lblume> it's only to do a benchmark, eg from a live cd, only one one half of the mirror[11:19:58] <kinesis> ok, understood. thanks![11:21:39] <lblume> what are the firmware revisions of those disks, BTW? I remember updating mine because there were apparently fairly good reasons to do so, even though Seagate didn't give much details.[11:29:45] <kinesis> dont know how to find it. Revision: CC46[11:30:29] <kinesis> its from format utility[11:31:45] <lblume> This one was rather quickly replaced by CC49.[11:32:19] <kinesis> you think i need to update them?[11:32:35] <lblume> No idea of the difference, though, but that they released a newer version so quickly is suspicious.[11:33:18] <lblume> Maybe worth a try. The upgrade path is rather painless, and with a mirror, your data are covered.[11:33:25] <kinesis> hmm, but in windows i have normal write speeds on them :-\[11:34:06] <kinesis> i will keep it in mind, thanks.[11:35:02] <lblume> I know. You might just be hitting a driver bug. Difficult to say, but well, Windows drivers have been known to work around firmware issues that impacted other systems.[11:36:26] <kinesis> also will try to check this system on other, non seagate disks.[11:40:12] *** merzo has joined #openindiana[11:47:51] *** InTheWings has joined #openindiana[11:51:56] *** Whoopsie has quit IRC[11:54:05] *** Whoopsie has joined #openindiana[11:54:05] *** ChanServ sets mode: +v Whoopsie[12:01:59] *** befreax has quit IRC[12:08:43] *** gea__ has joined #openindiana[12:15:19] <DontKnwMuch> share names in smb are all lowercase, is this settable somewere? you konw, pDataUsers is better than pdatausers[12:17:37] *** gea__ has quit IRC[12:17:37] *** SH0x has quit IRC[12:17:54] *** |AbsyntH| has quit IRC[12:22:45] *** SH0x has joined #openindiana[12:27:26] *** ianj has quit IRC[12:32:23] *** gervystar has quit IRC[12:33:02] *** held has quit IRC[12:34:57] *** bens1 has joined #openindiana[12:45:22] *** jwit has quit IRC[12:55:00] *** Naresh` has joined #openindiana[12:56:19] *** Naresh has quit IRC[12:58:43] *** DrLou has joined #openindiana[12:58:43] *** ChanServ sets mode: +o DrLou[13:05:58] <DontKnwMuch> does zfs list show compressed size or real size in its USED column[13:07:01] <nettezzaumana> DontKnwMuch: try it ..[13:07:06] *** merzo has quit IRC[13:07:53] *** pjama has joined #openindiana[13:08:28] <DontKnwMuch> seems like compressed, zpool list shows something else, right click on properties on share something else again...[13:09:30] *** held has joined #openindiana[13:12:09] *** McBofh has quit IRC[13:12:26] <nettezzaumana> honestly, i don't understand what good for is fs compression ... sounds to me like poor abilities of fat32 in mid 90's[13:12:37] *** mikw has quit IRC[13:14:12] <DontKnwMuch> compression is very good for compressable fs, I have 1TB of 2x compressed fs and this is great... no slowdown whatsoever, cpu utilisation is higher, but that is what it is for[13:15:46] <nettezzaumana> DontKnwMuch: i know how to use a basic tools like gzip, bzip2, zip or whate4ver else :P[13:16:12] <DontKnwMuch> sure, but why if the filesistem does it for you on the fly[13:16:36] <DontKnwMuch> and if 120.000 files are zipped in one zip... no good, right[13:18:44] <nettezzaumana> DontKnwMuch: unfortunately my porn collection consists yet from about 500 files but once i'll be so great collecter as you i'll reconsider it (eg using compression)[13:20:20] <DontKnwMuch> ahh... bit with the porn collection there will be no compression gain, except if you collect porn books and texts[13:21:22] <raichoo> \o/[13:22:33] *** xmikus01 has joined #openindiana[13:23:55] *** held has quit IRC[13:24:04] *** McBofh has joined #openindiana[13:24:28] *** mikw has joined #openindiana[13:26:25] *** xmikus01 has quit IRC[13:27:49] *** jwit has joined #openindiana[13:29:16] *** McBofh has quit IRC[13:38:16] *** |AbsyntH| has joined #openindiana[13:38:33] *** tsoome1 has joined #openindiana[13:39:51] <dkeav> he likes the articles[13:40:02] *** tsoome has quit IRC[13:40:22] *** tsoome has joined #openindiana[13:41:30] *** kohju has quit IRC[13:43:07] *** tsoome1 has quit IRC[13:43:57] *** kohju has joined #openindiana[13:54:16] *** McBofh has joined #openindiana[13:54:44] *** sivanov_ has joined #openindiana[13:56:02] <sivanov_> is there a way to limit kernel to certain cpu cores?[14:07:18] *** Whoopsie has quit IRC[14:13:47] *** tsoome has quit IRC[14:25:18] *** lkocman has quit IRC[14:27:23] *** dijenerate has quit IRC[14:29:02] *** geoff_ has quit IRC[14:31:06] *** freedomrun has joined #openindiana[14:31:40] *** McBofh has quit IRC[14:33:52] *** McBofh has joined #openindiana[14:35:55] *** tnorgaard has joined #openindiana[14:39:35] <sivanov_> intrstat shows some 130K interrupts on bge#0, it eats 50% of one core[14:39:46] <sivanov_> whats wrong ?[14:40:21] *** lkocman has joined #openindiana[14:40:26] <sivanov_> the interface is transmitting less than 100KB/s[14:40:38] <nettezzaumana> everything is wrong .. liberty is killed by hypocrites who roam in gangs all around[14:40:39] <tomww> switching to interrupt less mode for that interface...[14:40:47] <Triskelios> sivanov: this is post-update?[14:40:47] <tomww> if the ethernet chip supports that.[14:40:53] <Triskelios> sivanov: not a fresh install?[14:41:02] <tomww> other idea is disable sharing the interrupt with other hardware[14:41:29] <sivanov_> this is oi_148[14:41:38] <tomww> there is a command to display the current interrupt assignment (not at hand, others might have it)[14:41:46] <Triskelios> sivanov: did you update?[14:41:52] <sivanov_> nope[14:42:06] <Triskelios> sivanov: please do, there are serious issues in bge[14:42:24] <sivanov_> hmmm[14:42:33] *** mikw has quit IRC[14:42:39] <Triskelios> see also #544[14:43:58] <lblume> tomww: pcitool?[14:44:15] <sivanov_> i played with dladm set-linkprop -p cpus=0,1 bge0[14:44:25] <sivanov_> also with psradm[14:46:14] <Triskelios> sivanov: updating will fix it. I know because I accidentally introduced the bug[14:46:40] *** POloser has quit IRC[14:48:37] <sivanov_> also i see kernel message that int 19 is shared[14:49:00] *** McBofh has quit IRC[14:49:01] *** chrisridd has joined #openindiana[14:49:01] *** ChanServ sets mode: +v chrisridd[14:49:10] <dkeav> Triskelios: i think your mic is off or something[14:49:11] *** CoilDomain has joined #openindiana[14:50:28] <chrisridd> Can zones be created in oi_147? I'm getting an error on zoneadm install: "Sanity Check: Looking for 'entire' incorporation.[14:50:28] <chrisridd> Installing: Core System (output follows)[14:50:28] <chrisridd> pkg: No solution was found to satisfy constraints[14:50:28] <chrisridd> ERROR: failed to install package"[14:52:02] <sivanov_> do dladm set-linkprop setting stay accress reboots?[14:52:34] <sivanov_> Triskelios, so just do image-update?[14:53:08] *** McBofh has joined #openindiana[14:53:14] *** Naresh`` has joined #openindiana[14:55:45] *** russiane39 has quit IRC[14:55:51] *** mikw has joined #openindiana[14:55:54] <sivanov_> or update to oi_151?[14:57:22] *** Naresh` has quit IRC[15:00:17] *** sdward has joined #openindiana[15:02:15] *** mikw has quit IRC[15:02:31] *** mikw has joined #openindiana[15:03:14] *** mikw has quit IRC[15:03:35] *** tnorgaard_ has joined #openindiana[15:03:36] *** mikw has joined #openindiana[15:04:03] *** tnorgaard has quit IRC[15:04:03] *** tnorgaard_ is now known as tnorgaard[15:04:34] *** Naresh`` is now known as Naresh[15:04:37] *** Naresh has joined #openindiana[15:05:10] *** mnaser has joined #openindiana[15:05:52] *** mikw has quit IRC[15:06:11] *** mikw has joined #openindiana[15:07:10] *** mikw has quit IRC[15:07:26] *** mikw has joined #openindiana[15:09:55] <dkeav> no it won't upgrade to 151 unless you change your publisher uri[15:13:32] <sivanov_> ok, see the new bge after update[15:14:31] *** sivanov_ has quit IRC[15:18:02] <kinesis> anyone knows how to configure the system so that ACL is not deleted when writing files over nfs?[15:18:29] *** axisys has quit IRC[15:18:34] <kinesis> aclmode deleted now from zfs properties.[15:20:13] *** sdward has left #openindiana[15:21:09] <kinesis> i need ACL to be inherited from parent.[15:21:34] <kinesis> if i use cifs its ok, but nfs drop my acl's[15:23:04] *** freedomrun has quit IRC[15:23:27] *** freedomrun has joined #openindiana[15:23:41] *** sivanov_ has joined #openindiana[15:25:50] <sivanov_> apparently, i see an improvement: a 25GB file transfers@65MB/s from cifs share. was 35MB/s[15:28:14] <tomww> lblume: echo ::interrupts | mdb -k[15:28:22] <dkeav> nearly double is always an improvement[15:29:15] <sivanov_> now i get 4k interrupts from bge, transferring a file[15:31:37] <dkeav> all fixed up[15:31:56] <lblume> tomww: Looks quite like pcitool :-)[15:32:28] <tomww> all the same data source. so solaris kernel is a data mining tool.[15:32:44] <sivanov_> though, i dont understand why the ints are on cpu #3 while i configured them to run on #1,2[15:32:52] <sivanov_> *#0,1[15:33:20] <Wraithh> Interrupts on cpus 1+2 = 3 of course :)[15:34:11] <tomww> shared interrupts with other drivers are ususally problematic if you need troughput.[15:36:11] *** leoric has left #openindiana[15:36:48] <sivanov_> i remember it from tuning the NT: limit a NIC to certain CPU[15:37:16] <sivanov_> now i get little more MB/s[15:37:27] <sivanov_> limited to cpu #0[15:38:30] <tomww> can you share the output of the interrupt listing (e.g. pastebin, dpaste,...) ?[15:39:35] <sivanov_> but what interesting, interrupts are still running on #3[15:41:33] <DontKnwMuch> How can I mount a smb share, I have the smb client service running mount smbfs //192.168.0.2/share_name /path_to_mount_point does nothing[15:42:49] <sivanov_> tomww:http://www.pastie.org/2182950[15:43:07] <DontKnwMuch> oh. I forgot the -F[15:45:14] *** lkocman has quit IRC[15:49:53] *** skeeziks has joined #openindiana[15:50:39] *** bens1 has quit IRC[15:50:43] *** DerSaidin has quit IRC[15:51:22] <sivanov_> maybe i will get more MB/s if i switch to Jumbo frame[15:51:54] *** Whoopsie has joined #openindiana[15:51:54] *** ChanServ sets mode: +v Whoopsie[15:52:24] <sivanov_> however since the clients are windows 7s i expect that bad things may happen[15:53:10] *** mikw has quit IRC[15:55:56] *** DerSaidin has joined #openindiana[15:55:56] *** DerSaidin has joined #openindiana[15:56:20] <DontKnwMuch> you upgraded to 151?[15:56:29] <sivanov_> no[15:56:48] <sivanov_> just image-update[15:56:54] <DontKnwMuch> ah.. ok[15:57:03] <lblume> Windows supports jumbo frames just as well.[15:57:47] <viridari> Windows supported Jumbo frames before Sun fully did.[15:58:19] <sivanov_> at work me and my comrade broke network on win2008r2 playing with jumbo buffers[15:58:24] <viridari> back in the 1990's you could buy Sun servers with GigE interfaces and no jumbo frame support[15:59:48] <DontKnwMuch> jumbo frames has to be set on every machine to work, at least I found this out by breaking things...[16:00:16] *** lkocman has joined #openindiana[16:01:40][16:02:14] <DontKnwMuch> I asked as my echo ::interup.... does look completely different: and has a lot of dahses where irq are supposed to be: http://www.pastie.org/2183033[16:02:32] *** andygraybeal has quit IRC[16:03:27] <sivanov_> i bought some cheap FC hardware and now all my home computers boot from FC and enjoy the easy zfs snapshoots[16:03:56] *** smrt has quit IRC[16:04:15] *** smrt has joined #openindiana[16:05:06] *** miine has joined #openindiana[16:06:42] <yalu> cheap fc hardware? where, where?[16:08:06] *** Whoopsie has quit IRC[16:08:20] <sivanov_> ebay[16:08:41] <sivanov_> brocede 200e for $150[16:08:48] <sivanov_> *brocade[16:09:14] <sivanov_> qla2460 for $32-35, 6 pieces[16:09:36] <yalu> that's quite a deal[16:09:43] <sivanov_> and 10 kg of LC-LC patchcords for $15[16:09:59] <yalu> do they count it in kg these days :-)[16:10:04] <viridari> I just threw out boxes and boxes of FC patch cords yesterday[16:10:20] <yalu> new these things cost a fortune[16:10:29] <viridari> these were new in plastic still[16:10:35] <sivanov_> you can try and sell each for $5-10 :D[16:10:48] *** axisys has joined #openindiana[16:11:01] <viridari> yeah probably should have.[16:11:52] <sivanov_> saw brocade m4400 for $250 last week[16:12:10] <yalu> and storage cabinet?[16:12:12] <sivanov_> was bought fast to buyout though[16:12:18] *** freedomrun has quit IRC[16:14:01][16:21:12] <yalu> I wonder what it's like to have a FC storage box heating your home[16:22:43] <dkeav> quite nice in january[16:23:30] *** tnorgaard has quit IRC[16:24:33][16:25:13] <sivanov_> to warm up by it while drinking vodka[16:29:59] *** forquare has joined #openindiana[16:38:01] *** Botanic has joined #openindiana[16:38:01] *** Botanic has quit IRC[16:38:01] *** Botanic has joined #openindiana[16:41:35] <blues_> hmm[16:42:03] <blues_> got a little excited yesterday, ordered the sas controller and expander...but not sas-to-sata cables[16:42:19] *** held has joined #openindiana[16:42:32] *** Micr0mega has left #openindiana[16:42:37] <lblume> there is no sas to sata cable[16:43:32] <lblume> sas cables should fit on sata drives[16:43:50] <Wraithh> There are multilane sas cables[16:45:13] *** SH0x has quit IRC[16:45:44] <lblume> they are just cables, they can carry sas or sata information[16:46:19] <lblume> you can't «convert» them to sata[16:46:21] <blues_> i may be getting hte terminology wrong... i just know the cables that came with the expander are not made to fit a sata header[16:47:01] <lblume> They really should. Sata drives are able to connect to sas.[16:47:26] <lblume> But yes, maybe I'm not understandng correctly :-)[16:48:33] <dkeav> lblume: he means a mini-sas to 4x sas/sata[16:48:34] <blues_> http://images10.newegg.com/productimage/12-228-057-02.jpg[16:48:47] <blues_> both ends look like the end on the left in that pic[16:48:54] *** bens1 has joined #openindiana[16:49:50] <blues_> in any case, i don't have what i need... so whoops[16:50:00] *** tnorgaard has joined #openindiana[16:50:18] <lblume> Get it now :-)[16:50:49] <blues_> just so i sound like less of an idiot going forward, what i hold in my hand is a straight mini-sas cable, yes?[16:52:24] *** raichoo has quit IRC[16:53:07] <lblume> I merely nkow the SFF8088 and SFF8045 formats, sorry :-)[16:55:56] *** kinesis has quit IRC[16:56:31] *** kimc has joined #openindiana[17:02:03] *** syoyo_ has joined #openindiana[17:02:33] *** syoyo has joined #openindiana[17:09:22] *** tnorgaard has quit IRC[17:10:08] *** lkocman has quit IRC[17:13:38] *** tnorgaard has joined #openindiana[17:13:51] *** tnorgaard has quit IRC[17:17:02] *** chrisridd has left #openindiana[17:18:55] <dkeav> blues_: see that single minisas end in that pictures?[17:19:00] <dkeav> thats SFF8088[17:19:27] <blues_> what is the difference between 8088 and 8087 ?[17:20:02] <viridari> 1.[17:20:22] <dkeav> err sorry, picture IS 8087[17:20:27] <lblume> IIRC, 8088 is mostly for external use, and 8087 for internal.[17:20:31] <dkeav> 8088 is external[17:21:01] <lblume> So 8088 is made to not be easily pulled by mistake[17:21:18] <blues_> so here's the sucky part about living in TN... its not like i can run out to fry's and grab a few 8087 -> 4x sata cables[17:21:42] <dkeav> http://en.wikipedia.org/wiki/File:SFF_8088.jpg[17:21:50] <dkeav> http://en.wikipedia.org/wiki/File:SFF_8087.jpg[17:21:55] <dkeav> there![17:21:58] <dkeav> :D[17:22:00] <sivanov_> are ibm m1015 good sas hbas to use with solaris?[17:22:12] <lblume> Heh, I tend to order my SATA/SAS stuff in other countries, so I know the feeling :-D[17:22:26] <blues_> definitely not gonna confuse the two.[17:22:48] <dkeav> no, they are easy to distinguish[17:23:02] <sivanov_> from what i know i could buy them for like $60-70 and flash with IT mode firmware...[17:23:10] *** miine has quit IRC[17:23:40] <dkeav> sivanov_: find out what chipset they use[17:23:49] <dkeav> they are most likely rebadged products[17:24:11] <sivanov_> it seems like its lsisas2008[17:24:20] <sivanov_> sas 6gb[17:24:34] <blues_> i've seen some other people building S11E builds with them[17:25:19] <OniAtWork2> I cant get that in IL either[17:25:38] <blues_> Don't you have Fry's up there?[17:26:06] <OniAtWork2> There's one in Chicago, might as well be the moon for all the good it does me[17:26:25] <dkeav> OniAtWork2: what do you need?[17:26:29] <dkeav> i'm in illinois[17:26:44] <dkeav> and not chicago[17:26:45] <dkeav> :D[17:26:48] <blues_> I'm actually contemplating driving down to Fry's in Milton, GA just so i can get the cables to work on this[17:27:10] <dkeav> blues_: you're not a patient geek are you[17:27:15] <OniAtWork2> really, dkeav? where-abouts are you?[17:27:17] <dkeav> err thats an oxymoron isn't it[17:27:35] <blues_> i'm extremely patient until i'm ready to build[17:27:38] <dkeav> OniAtWork2: dunno how well you know illinois, so lets say c-u area[17:27:39] <blues_> then i want to get on with it[17:27:49] * OniAtWork2 is up in rantoul[17:27:52] <dkeav> ahh[17:28:03] <OniAtWork2> yes, I know central IL pretty well[17:28:05] <dkeav> i'm south of danville about 30 miles[17:28:09] <blues_> i've been speccing out this damn ESXi box for a month[17:29:16] <OniAtWork2> blues: please, please, please, check the HCL before you buy anything[17:29:31] <OniAtWork2> ESXi is REALLY picky about that.[17:29:36] <blues_> oniatwork2: everything is coming from the HCL[17:29:41] <blues_> thats why its taken so long[17:29:41] *** miine has joined #openindiana[17:29:41] <OniAtWork2> good[17:30:07] <blues_> going with a supermicro box, x5600 series xenon, still deciding on am't of ram[17:30:13] <blues_> no local storage[17:30:33] <dkeav> err besides boot drive[17:30:45] <blues_> i can actually get a supermicro box that has 10gigE and stay in budget[17:30:55] <blues_> gonna boot from flash[17:31:01] <dkeav> usb?[17:31:04] <blues_> yeah[17:31:07] <dkeav> ehhhh[17:31:10] <blues_> has internal connectors[17:31:24] *** alcir has quit IRC[17:31:31] <dkeav> try it, but try to make it as read-only as possible at some point[17:32:20] <dkeav> and check the 10gigE chipset in the HCL if its good to go, definatly worth getting now especially if you are looking for performance to a single client, eg nfs link[17:32:42] <lblume> usb performance of Solaris is rather less than stellar, not sure that OI would fare any better, does it?[17:32:51] <sivanov_> i used to boot a freebsd router from the flash and every month or so it would unable to write ufs journal to it and hang[17:32:52] <dkeav> not really[17:33:23] <blues_> the OI box will boot from a small ssd. ESXi will boot from the usb flash[17:33:30] <dkeav> you shouldn't have used journaling with a usb flash drive...[17:33:56] <OniAtWork2> I think all the supermicro stuff was full system certified, so a 10GE that comes in it should be fine[17:34:31] <dkeav> sivanov_: its better in that instance to use read-only UFS, with no softupdates or journalling, and mount as many things in memory device as possible[17:35:05] <dkeav> OniAtWork2: yea i know you are pretty much good to go with supermicro[17:35:06] <dkeav> BUT[17:35:13] <blues_> OniAtWork2: yeah, its system certified, but i'm still checking...[17:35:14] <dkeav> double checking never hurt before spending money ;)[17:35:42] <sivanov_> well, the usb boot was temporary, untill i find 30gb ssd for under $30:)[17:35:48] <dkeav> i know lots of things that are said to work with other things, oh but wait that was only for rev A.1, not A.2 or newer, sorry[17:35:52] <dkeav> grr[17:36:07] <dkeav> you need 30gb for a router?[17:36:16] <blues_> case in point: have ot make sure i get the K revision of whatever supermicro i get to make sure it can properly cool / power the 5600 series chip[17:36:21] <dkeav> no offense but you're doing it wrong[17:36:23] <sivanov_> its also runs rtorrent....[17:36:39] <dkeav> i'll stick to previous statement[17:37:25] <dkeav> why wouldn't you mount a network share for your torrents to save to[17:37:29] <blues_> 10gigE connecting host to san would be nice.. i was planning on iSCSI across 2 gigE ports[17:37:58] <dkeav> 2 gigE ports won't do you any good in that instance[17:38:02] <dkeav> other than redundancy[17:38:09] <dkeav> won't help with capacity[17:38:44] <sivanov_> unless the initiator fully supports mpio or mc/s[17:39:10] <blues_> i asked in the #vmware channel before going down that path... i won't get 2x 1gigE, but apparently i should get 1.4 ~ 1.6[17:39:33] <sivanov_> vmware can mpio, altho its different quistion how[17:41:14] <sivanov_> right now i have an infortrend a16e sitting at work waiting for resolution of a ticket @ infortrend regarding low performance with ESX on 4 gbe links[17:41:55] *** syoyo has quit IRC[17:42:08] <sivanov_> can only say, that s11e performs even worse[17:42:26] <sivanov_> actually much worse[17:50:01] *** |AbsyntH| has quit IRC[17:53:18] <blues_> hmm.. well.. if i can find some damn cables within a non-obscene drive, i can test multi-path iScsi between ESXi and OI this weekend[17:53:57] <OniAtWork2> blues_, I know supermicro doesn't cert their stand-alone mainboards. there are some they make that will kind-of work but just fall over under load[17:54:17] <blues_> Yeah, its the systems that are cert'd, not the boards[17:55:04] *** Cobi has quit IRC[17:56:18] <sivanov_> there is a company hare in RU, that sells their product called trinityNAS, supermicro based & nexenta with some patches. guys claim it does 1.8GB/s read 1.5GB/s write with 8 drives in raidz10[17:56:25] <sivanov_> *here[17:56:58] *** DanaG has joined #openindiana[17:57:11] *** DanaG has quit IRC[17:58:07] <dkeav> raidz1*[17:58:40] <sivanov_> you know what i mean, stripe of mirrors[17:58:56] <sivanov_> *raaid0 of mirrors[17:59:30] <dkeav> raid 10[17:59:33] *** forquare has quit IRC[17:59:36] <dkeav> no z[17:59:38] <dkeav> :D[18:00:10] <dkeav> sivanov_: i assume he is claiming that as a local read/write speed[18:00:24] <sivanov_> of course[18:00:48] <jkimball4> i feel like zfs is super slow on my hdd though i can't say if the hdd isn't just crap despite being fairly recently purcahsed[18:01:29] <Wraithh> Btw what's the current status of zfs with 4k block drives?[18:01:31] <sivanov_> zfs sure slows things down[18:01:57] <OniAtWork2> compression seems to help though[18:02:17] <sivanov_> Wraithh, afaik oi_148 should have pach[18:08:33] <dkeav> Wraithh: not great[18:08:38] <dkeav> can be made to work[18:09:00] <dkeav> at worst you suffer about a 10%ish hit to write performance[18:09:22] *** tsoome has joined #openindiana[18:09:29] <dkeav> if you can make a pool with ashift=12 they are supported[18:09:55] <dkeav> be warned that some expanders do not play well though, so its still best to use 512b disks[18:10:09] *** Cobi has joined #openindiana[18:11:12] *** sivanov_ has quit IRC[18:11:20] <miine> dkeav: don't know why people make a fuss about 10% loss of write performance...[18:11:46] <dkeav> yea its rather miniscule[18:11:47] <Wraithh> I heard about someone patching the zpool binary to adjust the alignment[18:11:50] <miine> dkeav: if it is so important - why use 2 or 3 TB consumer disks ??[18:11:58] <dkeav> exactly[18:12:04] <blues_> this should work fine shouldnt' it?[18:12:05] <blues_> http://www.amazon.com/Adaptec-2236600-R-Internal-Mini-SASX4-SFF-8087/dp/B000GU04X2/ref=sr_1_3?ie=UTF8&qid=1310141703&sr=8-3[18:12:21] <miine> dkeav: and even then in most cases ethernet will be the bottleneck...[18:12:33] <dkeav> i would be more concerned about the crappy discs and long TLER's[18:13:00] <dkeav> blues_: yes[18:13:52] *** SH0x has joined #openindiana[18:15:49] *** JT-EC has joined #openindiana[18:15:55] <miine> dkeav: do all disc on a single controller "starve" while one Non-Raid disc is recovering?[18:16:43] <miine> dkeav: I thought that soft-raid would behave different, but maybe I'm wrong and it's a controller implementation issue...[18:17:07] <tsoome> it depends on controller[18:17:20] <dkeav> if the controller kicks it out fast enough sure[18:18:19] <dkeav> but sometimes it can kick it out on you for a simple bad block, and you have a 2tb resilver going on or something, what happens if another of your cheap drives throws a long TLER and goes offline[18:18:25] <dkeav> not good[18:18:49] <dkeav> i'll pay the extra 20 bucks or whatever for decent drives[18:20:39] <miine> dkeav: hmm. which 2 TB drives are the decent ones?[18:20:51] <dkeav> consumer grade?[18:21:03] <Wraithh> WD 2TB Blacks are really nice[18:21:08] <dkeav> yes[18:21:24] <Wraithh> somewhat more expensive than some others though[18:21:26] <dkeav> they are basically the RE4's without TLER which is fine with zfs[18:23:02] *** hajma has quit IRC[18:23:09] <jkimball4> my pool says it has 11 gig of space left but zfs list only shows about 2 gig available for each filesystem. hint to limit the different?[18:23:41] <dkeav> parity?[18:23:57] <tsoome> zpool does show raw space[18:23:58] <jkimball4> yeah parity would be good :)[18:24:04] <dkeav> pool shows raw space[18:24:11] <dkeav> now add in parity and filesystem[18:24:29] <jkimball4> mm.. i see[18:24:37] <jkimball4> i forget about the overhead[18:28:27] <tsoome> that difference is a bit confusing:P[18:28:45] *** kforbz has joined #openindiana[18:29:59] <dkeav> can be[18:30:01] <blues_> god bless amazon and their saturday delivery[18:30:17] *** held has quit IRC[18:30:43] <dkeav> blues_: http://xkcd.net/921/ you tomorrow[18:30:48] * OniAtWork2 shudders about bad TLER[18:31:28] <blues_> lol. love the alt-text[18:32:40] <OniAtWork2> I dont know where they get the people to deliver on Sat, but I wish I did (I have a few uses for ninja)[18:35:05] <blues_> i'd rather pay 30 bucks for shipping than 60 + time and travel to go to fry's.. so woot[18:36:23] <OniAtWork2> well, good luck![18:40:04] *** kforbz has quit IRC[18:43:14] *** anickname has joined #openindiana[18:43:26] *** anickname is now known as randomnickname[18:43:43] *** randomnickname is now known as enterkey[18:44:31] <enterkey> How the heck do I disable that effin' system beep?! It's so loud that it's rendering OpenIndiana unusable![18:45:24] <viridari> find the speaker in your server case, follow the two wires to the system board and remove the jumper[18:45:42] <tsoome> there are very smart people in internetz, they did wrote entire libraries full of books about that. you can find all those books with google...[18:46:01] <quasi> isn't it a gnome thing? keyboard bell or something[18:46:01] <enterkey> tsoome: I read them all. 'xset b 0' does nothing.[18:46:10] <tsoome> then you didnt read them all[18:46:10] <enterkey> Well, AFAIK them all.[18:46:20] <enterkey> Editing GConf didn't seem to work, either.[18:46:23] <tsoome> because xset was only good for s10[18:46:36] <JT-EC> echo "set bell-style none" > .inputrc[18:46:49] <enterkey> JT-EC: set bell-style none does nothing, too.[18:46:57] <JT-EC> Check Volume Controls -> Preferences for machine beep option[18:47:07] <JT-EC> Disable beep in Terminal preferences[18:47:31] <JT-EC> Disable beep in about:config for Firefox[18:47:57] <JT-EC> inputrc does work. Did you log out and back in?[18:50:11] <enterkey> Ah...I'm an idiot. I didn't even consider logging out and back in.[18:50:17] <enterkey> Thanks![18:50:47] *** enterkey has quit IRC[18:51:03] *** GateKeeper has joined #openindiana[18:51:18] *** GateKeeper is now known as Guest81678[18:52:05] *** synegy34 has quit IRC[18:59:16] <tsoome> :D[18:59:23] <jkimball4> i usually mute it in the volume controls for gnome[18:59:36] <jkimball4> but inputrc will turn it off when i login after boot?[19:00:05] <jkimball4> is there such a thing as /etc/inputrc or just in $HOME?[19:00:19] *** raichoo has joined #openindiana[19:00:27] <tsoome> some day, when its full moon, the solaris will get sane defaults....[19:00:44] <jkimball4> that's funny :p[19:00:49] <tsoome> or openindiana:P[19:03:16] *** TPickle has joined #openindiana[19:04:39] *** held has joined #openindiana[19:23:11] *** bens1 has quit IRC[19:27:15] *** johnny has joined #openindiana[19:27:40] <wonko> tsoome: solaris has sane defaults, it's the rest of the world that's insane![19:27:44] <wonko> or something. :-D[19:28:03] <tsoome> :P[19:29:08] <wonko> or maybe i'm the one who's insane for having admined solaris for so long?[19:29:13] <wonko> that's the more likely option[19:29:32] <johnny> hi, I want to install freeradius. I am stuck at make: "make: Fatal error in reader: Make.inc, line xx". Steps so far: pkg install gcc-43; ln -s /usr/gcc/4.3/bin/gcc /usr/bin/gcc[19:30:01] *** johnny is now known as johnny_N30[19:31:19] <johnny_N30> i tried version from http://sunfreeware.com/ but i get: bash: /usr/local/sbin/radiusd: Invalid argument[19:35:20] <miine> johnny_N30: google desktop CBE .[19:35:49] <alanc> probably expecting you to run /usr/gnu/bin/make instead of /usr/bin/make or the other way around[19:35:58] <alanc> GNU & Solaris make are not 100% compatible[19:36:39] <johnny_N30> i'm running /usr/bin/make[19:36:41] <tsoome> usual gnu deathtrap...:P[19:37:20] <johnny_N30> compiling now, will see, thank you[19:37:46] *** gea__ has joined #openindiana[19:41:19] *** ivo_ has joined #openindiana[19:49:10] <johnny_N30> vanilla version from freeradius.org seem to compile well with gnu make[19:51:46] *** Reepicheep has joined #openindiana[19:52:49] *** sergefonville has joined #openindiana[19:53:59] <dkeav> not surprising[19:58:57] <wonko> foo, so S10 -> OI-151 via LU/BE isn't going to happen it looks like.[19:58:59] <wonko> dammit[20:00:27] <dkeav> would have been faster to just back and install[20:00:32] <dkeav> *back up[20:00:51] *** birvin has quit IRC[20:01:08] <tsoome> split mirror, destroy rpool, install OI on it and move data from old s10 rpool[20:01:20] <alanc> Live Upgrade can't even begin to contemplate how to convert a system from SVR4 to IPS pacakges[20:01:48] <wonko> for the machine at home, upgrade/install isn't a big deal[20:01:57] <wonko> but the box in colo is going to be a bit more complicated[20:02:09] <wonko> i might just have to suck it up and make a trip down there[20:02:14] <wonko> maybe upgrade the ram while i'm at it[20:02:34] <dkeav> scheduled downtime[20:04:34] <wonko> downtime isn't the issue[20:04:41] <wonko> the fact that it's really far away is[20:04:52] <miine> wonko: and?[20:04:59] <wonko> the only remote console i have is the serial port, so unless that magically works with the installer CD i'm SOL[20:05:53] <miine> wonko: want to know some magic?[20:06:04] <wonko> sure[20:06:48] <miine> wonko: build a usb stick based on live-cd. remove the stupid "language" / "keyboard" question. ALWAYS boot from the stick. then fast-reboot from the stick to the hd...[20:07:16] <miine> wonko: my idea. works great. KVM for cheap :-)[20:07:53] <wonko> hmmm, i'll have to play with that, that could be the secret to making the serial console work for install[20:07:57] <wonko> that would be awesome[20:08:08] <miine> wonko: they oi on the stick can even "phone home" to see if it should fast-reboot or not (in case something has to be corrected on the rpool)[20:08:39] <wonko> thanks for the tip, that's a nice way to deal with a stupid platform (PCs)[20:08:50] <miine> wonko: for install you could simply use distro constructor and change the grub menus setting to use serial console...[20:09:32] <miine> wonko: as those changes would not only apply to the live cd, but also to the install you would always have serial console access...[20:09:47] <wonko> yeah, i have grub set to use the serial console for the S10 install that's running on it now[20:09:55] <wonko> i don't remember if i had to do anything to solaris to make it use that though[20:10:35] <miine> wonko: but you're still f**** up if something went wrong with the rpool and need a reinstall or live medium... so my solution is really clever :D[20:11:09] <miine> wonko: and not to forget: CHEAP. 2 GB USB SD card (!! use them as they can be write protected) + adapter ~ 10 EUR ...[20:11:34] <wonko> yeah, i shall have to look into this fast-reboot thing[20:11:42] <wonko> this looks like it'll be the solution to my problems[20:12:32] <miine> wonko: yeah. it simply doesn't matter if the boot needs 2 minutes more or not.[20:13:01] <wonko> yeah, this isn't a mission critical machine anymore[20:13:06] <wonko> i've moved mail to google apps[20:13:06] <miine> wonko: still have to check out if via fast reboot the rpool could be raid-z ... that would be too cool...[20:13:17] <wonko> no need for me[20:13:24] <wonko> machine has 2x 73GB SAS and 2x 500GB SATA[20:13:36] <wonko> so raidz wouldn't help me anyway[20:13:37] <wonko> :-D[20:14:06] <yalu> hi, hello, how can I configure my system so I can log in as root? I want to rsync as root using passwordless ssh (I set permitrootlogin in sshd_config)[20:15:03] <tsoome> dont.[20:15:06] <dkeav> >.<[20:15:07] <miine> yalu: bad idea. put the sudo into the authorized keys file command and use a different user...[20:15:12] <dkeav> you're fired[20:15:32] <tsoome> build it to use normal user and get access. as told[20:16:08] <tsoome> root access over network is *bad* habit[20:16:30] <dkeav> should be a zero tolerance offense[20:16:33] <dkeav> imho[20:16:48] <tsoome> agree.[20:16:50] <yalu> so people keep telling me, but to be fair I never understood why. especially if you do it passwordlessly[20:17:11] <tsoome> till next ssh bug? :P[20:17:44] <yalu> could well be in the part were you say to disallow root access :-)[20:17:56] <dkeav> till someone gets your passwordless key files, ssh exploit, the fact that using proper users/groups/shells and permissions should be admin 101[20:17:56] <tsoome> any kind of privileged user exposure to net is potentail security issue[20:18:33] <yalu> true. but then again, if a cracker knows my username he can still execute sudo remotely (or whatever it is you do)[20:19:08] <tsoome> yes, but he does know root username for sure[20:19:08] <dkeav> o.0[20:20:16] <yalu> getting to know someones username is NOT hard. at work it's my mail adress.[20:20:56] <tsoome> anyhow, its same thing as not using root for normal user tasks[20:22:30] <yalu> well that's about using too much gunpowder really :-)[20:22:44] *** hajma has joined #openindiana[20:22:55] <tsoome> :)[20:23:34] <tsoome> it did took some time to remove even ssh root@host habit from our admins...[20:24:07] <tsoome> and some still do violate it when they can;)[20:24:53] <dkeav> if they want fired![20:25:11] <tsoome> our = our company[20:25:17] <yalu> suppose you really trust passwordless login and assume ssh only knows one remote hole in the default install in so many years... then your account is better protected with that method than with a remote sudo and a user that CAN log in remotely with a password, even if the pwd is as strong as root's[20:25:29] <yalu> personaly I have to admit I do it on a daily basis[20:25:59] <tsoome> noone is expecting the spanish inquisition![20:26:06] <dkeav> heh[20:27:34] <yalu> well I won't doubt that it's a security risk, which is what security is all about: risks. and there are much greater risks for the average server.[20:27:41] <tsoome> tbh, i think the root should be removed most of the permissions just as its done in trusted solaris:P[20:29:30] <yalu> hmmm not really a clue what you are talking about :)[20:29:32] <miine> oh. when talking about security: some services could get the certificates key from another service. anybody here knowing how that is implemented? pipes?[20:29:58] <tsoome> pkcs#11[20:31:22] <miine> tsoome: so there is no easy way to have a service inside a zone asking the global zone for the key...[20:31:36] <tsoome> no, not at all[20:32:14] <tsoome> local zone is another host[20:32:54] *** Guest81678 has quit IRC[20:34:02] <miine> tsoome: but the global zone could zlogin into it and have a pipe to a process which then could give the global zone's connected process the pid or service id of the requesting process/service. this could be checked if ok and if so the key could be transfered via pipe...[20:35:23] <miine> tsoome: this should work at least up to the key transfer as the global zone can see all pid's etc... . so if the key transfer would via pipe to the requesting service it would be relative easy to implement...[20:39:04] <tsoome> or implement some nice key exchange protocol over ip:)[20:39:57] <miine> tsoome: nope. that wouldn't be more secure as you have to check if the requesting process is the "right one"...[20:41:24] <miine> tsoome: the global zone could also check if the process is executed from the correct binaries (or even more fancy stuff I don't know to implement yet ;-) )...[20:41:49] <tsoome> zlogin is definitely quite ok;)[20:41:49] <miine> tsoome: if it seems hacked, shutdown the zone, rollback to known state and case closed :-)[20:42:48] *** merzo has joined #openindiana[20:42:49] <miine> in worst case there is then a DoS attack, but better then compromised certificates...[20:43:24] <miine> and we know - by seeing that the zone was rolled back - that there IS a security issue in the specific application...[20:44:21] <miine> would love to have somebody paying me to implement such stuff...[20:44:34] *** kart_ has quit IRC[20:45:01] <tsoome> :D[20:46:31] <miine> there is so much technology unused in oi / solaris which could be used in different ways so that linux etc. would be left in the dust...[20:47:34] <miine> its the best operation system i've seen so far. the only drawback is that's pain to use/learn...[20:50:08] <tsoome> ?[20:50:45] <tsoome> i'm in pain every time i have to touch some linux...[20:51:02] <tsoome> like holy water on vampire:P[20:51:28] <yalu> well, I basically grew up on Linux[20:51:46] <miine> I think there is much more, better and broader documentation on linux.[20:51:59] <tsoome> ?[20:52:02] <yalu> the community is different...[20:52:51] <miine> on oi / osol / solaris its more like collecting the stuff from different places yourself. and stuff is sometimes "snippets" refering to other snippets knowbody knowing if its right or not...[20:53:14] <richlowe> There's more documentation because there are more users[20:53:38] <miine> as I come from MacOS X, it was quit a shock to see how much time I had to spend just to have a working compile/package cycle...[20:53:40] <richlowe> in general, our manual pages are much better, everything else isn't. (excepting the DTrace and Linker and Libraries guides, which are spectacular)[20:53:46] <tsoome> go to oracle docs and use all the books from s11 section.[20:55:13] <miine> the thing is "cultural": suns business was to have you evaluate the software for free, but you were not allowed to use the gained knowledge within your business. to do that you had to go to their courses...[20:55:23] <miine> .. and pay at lot of $$.[20:55:49] *** DanaG has joined #openindiana[20:55:59] <DanaG> say, where did gcc4 go? It's disappeared from the repos.[20:56:31] *** axisys has quit IRC[20:56:45] <miine> DanaG: coming back from FreeBSD ? :D[20:57:36] <dkeav> whats wrong freebsd[20:58:30] <DanaG> Well, I'm still considering OpenSolaris...[20:58:38] <miine> dkeav: nothing. but DanaG had some unsolved oi issues yesterday and was tempted to stay at FreeBSD...[20:58:45] <tsoome> well, if you are skipping oracle docs because or some cultural belief, you may need to rethink your reasoning:P[20:59:00] <miine> DanaG: developer/gcc/gcc-432 is still there. pkg list -a | grep gcc ...[20:59:37] <miine> tsoome: the docs are ok. but the assume some decent knowledge.[21:00:02] <dkeav> speaking of freebsd, there is some stellar docs[21:00:03] <miine> tsoome: what I described was the license to some of Suns software.[21:00:15] <DanaG> Creating Planpkg: The following pattern(s) did not match any packages in the current catalog.[21:00:15] <DanaG> Try relaxing the pattern, refreshing and/or examining the catalogs: developer/gcc/gcc-43[21:00:28] *** kytibe has quit IRC[21:00:52] <miine> DanaG: then developer/gcc/gcc-43 ?[21:00:57] <DanaG> that "pkg" needs a leading space and colon. Otherwise you get Planpkg.[21:01:14] <DanaG> Same message for the full path.[21:01:47] *** Naresh has quit IRC[21:03:30] *** dws6045 has joined #openindiana[21:05:33] *** kytibe has joined #openindiana[21:06:22] <dkeav> http://pkg.openindiana.org/legacy/info/0/developer%2Fgcc%2Fgcc-43%404.3.3%2C5.11-0.133%3A20100306T002008Z[21:06:31] *** axisys has joined #openindiana[21:09:44] *** hajma has quit IRC[21:11:02] *** hajma has joined #openindiana[21:12:03] *** hajma has quit IRC[21:12:09] <DanaG> hmm, I wonder why it's not seeing that package.[21:12:21] <DanaG> It saw it fine yesterday (on a different drive).[21:12:34] <dkeav> legacy repo enabled?[21:13:42] *** sponix has quit IRC[21:14:18] *** hajma has joined #openindiana[21:15:50] <DanaG> both are enabled and sticky:[21:15:50] <DanaG> http://pkg.openindiana.org/dev/ above http://pkg.openindiana.org/legacy/[21:15:52] <herzen> why use an old gcc-4?[21:16:02] <herzen> http://staticdev.uk.openindiana.org:10003/info/0/pkg%3A%2F%2Fsfebuild%2Fruntime%2Fgcc%404.5.3%2C5.11-0.151%3A20110629T232855Z[21:16:21] <DanaG> Well, I'm not getting any matches when searching for gcc-4.[21:16:54] <dkeav> using pkg search?[21:17:08] <herzen> pkg set-publisher -g http://staticdev.uk.openindiana.org:10003 sfebuild[21:17:24] <herzen> that should make runtime/gcc accessible, I think[21:17:31] <DanaG> Anything local to US?[21:18:00] <herzen> this repo is a work in progress and not mirrored yet[21:18:22] <DanaG> ah.[21:18:49] <herzen> and it won't be getting released under that name, anyway[21:20:15] <DanaG> I wonder why the old one disappeared.[21:20:49] *** Botanic has quit IRC[21:23:54] <herzen> well, Oracle never really moved beyond gcc-3. and the OI devs have not yet settled on their own modern gcc package.[21:24:08] <herzen> that's the background, but I don't know why it disappeared[21:24:58] *** sponix has joined #openindiana[21:28:09] *** CoilDomain has quit IRC[21:30:37] <DanaG> weird... the 43 package is back.[21:35:34] <DanaG> hmm, napp-it installer seems to do nothing.[21:37:09] <DanaG> oh, it echoes into a log.[21:39:03] *** Botanic has joined #openindiana[21:39:03] *** Botanic has quit IRC[21:39:03] *** Botanic has joined #openindiana[21:40:24] *** geoff_ has joined #openindiana[21:43:24] <miine> DanaG: why use napp-it at all[21:47:39] *** freedomrun has joined #openindiana[21:48:53] <DrLou> Hey RoyK, or other ZFS mavens: How come a zpool scrub suddenly is VERY slow - as in 3K scanned in about 10 hours... ?[21:49:16] <RoyK> does iostat -en report errors?[21:49:28] <DrLou> and zpool status reports nothing![21:49:50] <DrLou> no errors at all![21:50:09] <DrLou> The only error is mine - that I've tried to get VirtualBox running![21:50:45] * RoyK wonders how that could do anything to the pool[21:51:01] <miine> at least it installs some device drivers.[21:51:11] <DrLou> I dunno - it's pretty weak evidence, I'll grant you - except that it's the BIG variable in the equation...[21:51:15] <miine> DrLou: everything else runs at normal speed?[21:51:45] <DrLou> other things are freaking out a little bit, as well, and acting weird, like segfaults.[21:52:03] <DrLou> shall I zpool scrub -s to cancel current?[21:52:05] <miine> DrLou: rebooted already?[21:52:13] <DrLou> no, waiting for sage advice[21:52:26] <DrLou> in fact, machine will not respond to GUI reboot command.[21:52:50] <miine> what about the interrupts?[21:52:52] <DrLou> is Vbox horrifically 'leaky'??[21:53:03] <DrLou> how to report them to you?[21:53:29] <miine> hmm. would have to consult the osol bible for that...[21:53:53] <miine> man intrstat ...[21:54:45] <RoyK> DrLou: try to stop the vbox processes[21:55:27] <DrLou> That's a bit funny in and of itself - Have 'quit' Vbox, yet its window remains open. Not good![21:56:12] <miine> kill the whole window manager / gnome or login ...[21:56:38] <RoyK> DrLou: aren't there some services running?[21:56:43] <DrLou> now complete mouse/keyboard freeze. I mean Siberia![21:56:45] <RoyK> svcs -a | grep -i v.*box[21:57:03] <DrLou> Now it's gotta be a hard reboot, I fear.[21:57:09] <miine> which vbox version?[21:57:52] * RoyK just upgraded his phone to Cyanogenmod 7 - whee![21:58:38] <DrLou> miine: Latest as of a week ago - 4.10 or something?[21:59:29] <miine> hmm. had issues with that (couldn't install the extension pack via gui but didn't try via cli...).[21:59:38] <DrLou> miine: do you have any advice about the write-through cache setting on VBox, with Oi the host?[21:59:59] <DrLou> I had no trouble installing extension pack, fwiw, using the VBoxManager[22:00:15] <DrLou> (never done did it that way before...)[22:00:32] <miine> DrLou: it's more an issue for oi as client I think...[22:01:17] <DrLou> That I'm not doing - am trying to run Linux as client on Oi host, to manage the few (damned) apps we still have which are built for Linux.[22:01:41] <miine> also there are some serious network issues in vbox. they are somehow releated to machine / oi / vbox version's. so it's getting a nice matrix whats works and whats not working...[22:02:11] <DrLou> My troubles never seem to be related to net, luckily.[22:02:34] <DrLou> Have had troubles with storage integrity _both_ times I've given VBox a real spin.[22:02:43] <miine> DrLou: then it should "just work" assuming that you do not access raw partitions as vmdks ...[22:02:45] <DrLou> What settings on that cache do you prefer?[22:02:51] *** tsukasa has joined #openindiana[22:02:57] <DrLou> no raw partitions in my case.[22:03:18] <RoyK> DrLou: did you try to check if there are any vbox services running?[22:03:28] <jkimball4> is there a way to tell what pools exist on a device?[22:03:42] <RoyK> jkimball4: zpool import[22:03:51] <miine> DrLou: no host i/o cache works for me...[22:03:57] <DrLou> RoyK: had to cold reboot -[22:04:12] <jkimball4> RoyK: ah didn't realize it could be used without a pool name for that :p[22:04:13] <RoyK> k[22:04:14] <jkimball4> silly me[22:04:16] <DrLou> miine: which is what the docs seem to really be pushing - and is the default, no?[22:04:28] <jkimball4> thanks that worked swimmingly[22:04:52] <miine> DrLou: yep. it's the default i think...[22:05:30] <miine> DrLou: oi 148?[22:06:12] <DrLou> yes, for the moment[22:06:32] <RoyK> anyone that knows an ETA for 151?[22:06:48] <miine> strange. but I remember having seen something like that too. maybe you should choose 4.04 or 3.x ...[22:07:22] <jkimball4> RoyK: you should know by now[22:07:25] <jkimball4> "soon" :)[22:07:26] <DrLou> but it's the 'like that' I don't understand - can't even tell exactly what's happening.[22:08:07] <johnny_N30> what tweaks should i consider when installing oi on USB 4gb thumb-drive? I already moved home dirs, swap files off drive[22:08:11] *** freedomrun has quit IRC[22:08:20] *** PMT has quit IRC[22:08:24] <miine> DrLou: dtrace would tell you ... but it won't help because you can't change it anyway other then trying which vbox versions works or not...[22:09:31] <DrLou> You mean dtrace 'inline' with the execution? Doesn't it work best with the dtrace interface code built in? (Or is it?)[22:09:34] <miine> johnny_N30: you will work on it?[22:10:25] <johnny_N30> headless server, there will be ZFS with 3 hard drives for data[22:11:03] <DrLou> Now, who wants to bet whether VBox will start up that snapshot without error?[22:11:07] <miine> johnny_N30: you will just dd the zpool into the stick?[22:11:45] <miine> johnny_N30: or did you use the live image?[22:11:58] <RoyK> johnny_N30: if you have sufficient memory, OI normally won't use swap. for the homedirs, use auto_home to map that elsewhere[22:12:36] <DrLou> miine: As I suspected, that snapshot reboots with root filesystem errors.[22:12:39] * RoyK guesses johnny_N30 just installed on the stick as if it were a drive[22:12:50] *** PMT has joined #openindiana[22:13:14] <johnny_N30> i'm installing it onto usb[22:13:18] <johnny_N30> yes[22:13:56] <RoyK> johnny_N30: should work well if it's just for data storage, even with swap and homedirs and logs on the stick[22:15:27] <RoyK> johnny_N30: I once broke an upgrade with moving /usr/mysql off the rpool (the upgrade created a new BE and upgraded there, and then the old mysql installation wasn't upgraded, which led to a few issues)[22:15:34] <johnny_N30> oh yeah, i'll move logs too[22:15:56] <RoyK> you shouldn't need to[22:15:59] <DrLou> So, what to do? Should I assume this is a Linux issue? (What is SuSE using - reiserfs still?) - or is it a VirtualBox issue? Or...?[22:16:03] <RoyK> unless the usb stick is very small[22:16:32] <miine> DrLou: its a vbox issue. otherwise oi wouldn't be affected...[22:16:39] <johnny_N30> i'm currently running nexenta 3 but it's no longer supported so I had to switch somewhere. I really like zfs[22:16:44] <RoyK> DrLou: it's probably not a linux issue (afaict), although using killer^Wreiserfs isn't among my favourites[22:17:03] <RoyK> johnny_N30: how big is this usb stick?[22:17:07] <miine> reiser was ruled guilty?[22:17:20] <RoyK> he's serving prison as of now, yes[22:17:20] <johnny_N30> 4gb[22:17:38] <johnny_N30> is this ok for headless server?[22:17:53] <DrLou> johnny_N30: ZFS is a joy.[22:18:10] <DrLou> Herr Reiser oughta be strung up....[22:18:44] <DrLou> honestly, don't know if it's Linux itself, or the VBox wrapper. Problem now is: How to diagnose? How to use the stuff?[22:19:07] <miine> DrLou: try latest vbox 3[22:19:17] <DanaG> No version of runtime/gcc-46 can be installed:[22:19:17] <DanaG> pkg://sfebuild/runtime/gcc-46 at 4 dot 6.0,5.11-0.151:20110629T234655Z: Suitable required dependency pkg:/system/library at 0 dot 5.11,5.11-0.151 cannot be found[22:19:44] <miine> DanaG: no wonder when you[22:19:51] <miine> are using oi 148...[22:20:33] <DanaG> Well, 4.3 isn't showing for me.[22:21:03] <miine> DanaG: and 4.6 is for oi 151...[22:21:17] <RoyK> johnny_N30: should be fine[22:21:18] <miine> DanaG: if you want to use 4.6 maybe you should build yourself...[22:21:31] <DanaG> I don't care what 4.x, as long as it's SOME 4.x.[22:21:39] <RoyK> johnny_N30: and for the home dirs and logs and so on, I wouldn't worry too much about them[22:21:44] <DanaG> er, wait, now the 4.3 is back?[22:21:55] <DanaG> Nope, no versions available.[22:22:46] <miine> DanaG: google for oi desktop cbe and use that stuff to create your build environment[22:26:30] <DanaG> Is the repo just in flux, or something?[22:27:45] <johnny_N30> RoyK: ok, thanks, will see how it goes[22:28:29] <RoyK> johnny_N30: if you have a nexenta zpool already, it should import without much problems[22:29:42] <johnny_N30> RoyK: you mean the data that I have on those 3 hard drives? I hope so, I am doing the backup now nonetheless[22:29:51] <DrLou> DanaG: yes, big discussions about gcc 4.6 going on internally. Watch this space.[22:30:08] <RoyK> johnny_N30: but did you say nexenta 3 wasn't supported any longer?[22:30:36] *** synegy34 has joined #openindiana[22:30:50] <johnny_N30> i think the community kind of died[22:30:58] <RoyK> oh[22:32:33] <RoyK> DrLou: after the reboot, to do you still have issues?[22:32:48] <johnny_N30> I hope I will grasp oi concept[22:33:08] <RoyK> it's not that hard if you know unices in general[22:33:53] <DrLou> After reboot, no. In fact, it had been running reliably for about a week until today.[22:34:06] <DrLou> I just now question reliability.[22:34:08] <RoyK> and the unix rosetta stone is quite handy http://bhami.com/rosetta.html[22:34:34] <RoyK> DrLou: what controller(s)?[22:34:57] <DrLou> You of all people know the controllers in this infernal thing - remember your script?[22:35:29] <johnny_N30> well it's only my hobby, I was using nexenta for openvpn, freeradius, postgresql, mysql, squid and webserver. I have to migrate all these services to oi now. It's very time consuming :([22:35:36] <RoyK> DrLou: sure, 9201s?[22:36:01] <RoyK> johnny_N30: both postgresql and mysql?[22:36:08] <johnny_N30> yes[22:36:28] <RoyK> mysql is in the repos, but IIRC postgresql isn't[22:36:32] <johnny_N30> mysql is a must for web-apps unfortunately[22:36:55] <johnny_N30> postgresql is in repos, i have 8.4[22:37:05] <RoyK> ok[22:37:07] <DrLou> I'm building postgresql for the repos - not to worry[22:37:22] <RoyK> just remember what I said about moving /usr/mysql off the root - it'll probably break upgrades[22:37:26] <johnny_N30> really? that's great, thanks DrLou[22:37:34] <DrLou> 9.n and everything after, with bells and whistles.[22:37:59] <RoyK> DrLou: does it come with blue LEDs?[22:38:08] <DrLou> We live and die by Postg reSQL[22:38:14] <tsoome> can i has blue leds?[22:38:23] <DrLou> for you, Roy, it will have the spinny, shiny things.[22:38:32] <RoyK> lol[22:38:42] <tsoome> my precioussss![22:39:13] <DrLou> I am building in a feature which will intermittently tell the developer how smart and good-looking he is, too.[22:41:09] <DrLou> RoyK: On this machine, if you recall, we've got the Intel Cougar Point controller working great - this is a P67A board.[22:41:38] <DrLou> Problem was with the Marvell 88SE9128 controller, which we cannot use at all.[22:41:47] <RoyK> no, I didn't remember ... I don't keep lists of other people's hardware - sorry :P[22:41:58] <DrLou> And here I had assumed...[22:42:39] <DrLou> You were interested, though, because your diagnostic script reported nothing, even on the Intel controller.[22:43:11] <johnny_N30> talking about controllers. Is asus u3s6 supported?[22:43:14] <RoyK> there's a parsing issue in that script[22:43:29] <RoyK> haven't gotten around to fixing it (yet)[22:44:22] <johnny_N30> 88SE9123[22:47:39] <DrLou> Johnny - got anything connected to that?[22:48:06] <johnny_N30> no, thinking of getting one, if it works with solaris[22:48:21] <johnny_N30> last time i checked it wasn't supported[22:48:42] <DrLou> I think I'm gonna go with 'NO' so far - I haven't been able to get ours working at all.[22:49:01] <johnny_N30> neither usb 3.0 nor sata 6gbs[22:50:58] <dkeav> yay! beer thirty[22:51:00] <dkeav> taa[22:52:25] <RoyK> johnny_N30: SAS2 controllers are supported, though[22:52:32] <RoyK> works well with SATA drives[22:52:52] <johnny_N30> but they are pricey :/[22:53:20] <RoyK> but then, the climb from 3 to 6Gbps wonæt help much performance-wise unless you're using SSDs or a SAS expander[22:53:49] <johnny_N30> i know[22:53:52] <RoyK> for direct attach, 3Gbps is well enough for spinning rust[22:55:31] <johnny_N30> hehe, got some more spinners laying around and wanted to make raidz2. I guess I'm staying with raidz.[22:56:58] <johnny_N30> Also if I'll spot some cheap controller I'll add ssd cache drivers but that's not priority now[22:58:02] <RoyK> if those databases are demanding, you should consider striped mirrors over raidzn[22:58:50] <tsoome> you can check with iopattern DTT script[22:58:54] <RoyK> this one performs rather well :) http://paste.ubuntu.com/640366/[22:59:06] <johnny_N30> they are not :) mostly just one user[22:59:15] <RoyK> ah[22:59:15] <RoyK> ok[23:03:59] *** Gnu_Raiz has quit IRC[23:04:57] *** ianj has joined #openindiana[23:09:27] <johnny_N30> wow, just saw that pastebin. that is nice[23:10:10] *** bens1 has joined #openindiana[23:11:52] <RoyK> C300s for L2ARC and a pair of OCZ Vertex 3 for the SLOG[23:12:14] <RoyK> pretty cheap too :)[23:13:29] *** bens1 has quit IRC[23:13:30] <johnny_N30> i think my sarcasm meter is broken. What controller and mobo?[23:15:12] <RoyK> Supermicro X8DTH-i/6/iF/6F (don't remember correctly, but that's what prtdiag told me), controller is LSI 3081[23:15:32] <RoyK> Hitachi deskstar drives for the data[23:15:36] <RoyK> 1TB[23:16:01] <RoyK> box cost about NOK 40k[23:17:14] <johnny_N30> mine is made out of abandon hardware :)[23:19:07] <RoyK> this is a file server, so I bought it for that purpose[23:19:25] <johnny_N30> how many users can this serve?[23:19:34] <RoyK> still rather hilariously cheap compared to other stuff we have[23:19:39] *** synegy34 has quit IRC[23:19:54] <RoyK> it can probably serve quite a lot more than those 30-ish users at that office[23:20:24] <RoyK> though some of them are rather tough on it, running met models etc from their PCs[23:20:48] *** axisys has quit IRC[23:23:59] <johnny_N30> one thing that's great about solaris is you can build powerful machines out of cheap (or relatively cheap) parts. I imagine what machine like that would cost if it was a brand name.[23:25:53] <RoyK> this is a supermicro box[23:26:12] <RoyK> but then, supermicro is rather cheap and we have a rather good deal with the local importer[23:27:32] <johnny_N30> supermicro is great value[23:29:50] *** skeeziks has quit IRC[23:30:41] *** hajma_ has joined #openindiana[23:31:53] *** hajma has quit IRC[23:34:34] *** Reepicheep has quit IRC[23:48:40] *** pothos has quit IRC[23:48:56] *** pothos_ has joined #openindiana[23:49:16] *** pothos_ is now known as pothos[23:52:03] <DontKnwMuch> this is my poor thing, supermicro also, 3tb hitachi, 2x sas2008 http://pastie.org/2185078[23:52:32] <DontKnwMuch> this cost me 3k[23:52:55] <RoyK> doesn't look that poor to me :)[23:53:13] <RoyK> but you might want some L2ARC for good performance during scrub/resilver[23:53:37] <RoyK> what sort of SSDs are you using?[23:54:20] <DontKnwMuch> F40 corsair: http://blog.corsair.com/?p=3720[23:54:42] <DontKnwMuch> these were cheap, I just hope they will survive a year or so[23:54:57] <RoyK> how much memory do you have in the box? and how is the write load?[23:55:55] <RoyK> 40GB for SLOG alone is sufficient for a machine with 80GB RAM and LOTS of writes[23:56:04] <RoyK> 5-10GB should be sufficient for most loads[23:56:15] <DontKnwMuch> 16GB ram, I would have 32Gb, but there are no 8GB ECC non registered ram modules available, its a X9SCA-F maiboard[23:56:40] <DontKnwMuch> I have no load, I am copying from old freenas thing, 4T to go...[23:56:42] <RoyK> with 16GB RAM, OI will use at most 8GB for SLOG[23:57:01] <RoyK> so better slice those up and use the leftovers for L2ARC[23:57:16] <DontKnwMuch> Is it ok to slice them, I really do not know[23:57:24] <RoyK> run format[23:57:28] <RoyK> partition[23:57:29] <RoyK> etc[23:57:43] <DontKnwMuch> I was 'told' this is not good, sortof[23:57:45] <RoyK> make sure to remove the slog from the pool first[23:57:59] <RoyK> it's not perfect, but better than no l2arc[23:58:09] <DanaG> DrLou: I wish they hadn't removed 4.3... I need it to compile Deluge.[23:58:40] <DontKnwMuch> I sortof thought that ram if there is enough of it makes l2arc unnecesary[23:58:56] <DontKnwMuch> sorotof[23:59:01] <RoyK> DontKnwMuch: l2arc helps a LOT during scrub/resilver[23:59:20] <DanaG> Say, if I put l2arc on an SSD, will it kill it rapidly?[23:59:55] <RoyK> DanaG: not really, they'll wear out, but that's just how they are