July 11, 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 11, 2011 [00:03:36] *** Botanic has quit IRC[00:03:50] *** Botanic has joined #openindiana[00:03:51] *** Botanic has joined #openindiana[00:08:31] *** systemonkey has quit IRC[00:10:33] *** Botanic has quit IRC[00:16:31] *** Botanic has joined #openindiana[00:22:40] *** raichoo has quit IRC[00:37:12] <DontKnwMuch> is anyone sucesfully "restore" a DC from OI snapshot. How old can it be? And what does a snapshot do with a large file (100GB) in NFS share which is in use by ESXI?[00:37:42] <DontKnwMuch> Is such file able to boot?[00:43:28] *** gea_ has quit IRC[00:44:04] *** gea_ has joined #openindiana[00:47:04] *** DeanoC_ has joined #openindiana[00:47:05] *** ChanServ sets mode: +o DeanoC_[00:48:04] *** DeanoC_ has quit IRC[00:48:44] <gea__> a ZFS snapshot is like a sudden power loss. I expect a DC controller to restart from a snap without problem, just do a restore of the VM folder[00:48:46] <gea__> If you need real hot snap, you must do a RESXi snap priot the ZFS snap.[00:50:10] *** DeanoC has quit IRC[00:51:45] <gea__> You can then go back to this state (with system and memory state) within ESXi[01:00:45] <sergefonville> how come, a unix socket does not perform awful compared to a TCP socket[01:01:10] <sergefonville> since, it seems to me that a unix socket, being a file can only be used by one process at a time, serving only one request[01:01:21] <sergefonville> or is that a false assumption?[01:01:48] <alanc> unix sockets aren't really files, they just appear in the filesystem name space[01:02:13] <sergefonville> still, are they for example threaded?[01:02:25] <sergefonville> since, google does not seem to know either :([01:02:27] <alanc> they usually perform much better than TCP sockets for local connections, since they don't have to go through all the TCP protocol overhead[01:04:03] <alanc> the socket is a communications mechanism, whatever program is using it could be multithreaded if it wants to be[01:04:11] <sergefonville> I'm testing with php and nginx, and am still nog sure, what I should actually test to determine what is bettter[01:04:45] <sergefonville> so if the process behind the socket is threaded, so is the socket, otherwise its fifo?[01:04:46] *** gea__ has quit IRC[01:06:11] <alanc> the question doesn't make sense[01:06:18] <alanc> sockets are always fifo[01:06:25] <alanc> Unix or TCP[01:07:06] <sergefonville> why, since one request can take longer to process than another, right?[01:07:17] <sergefonville> so a request started later, can finish earlier?[01:07:29] <sergefonville> or do all requests have to wait in line?[01:07:49] <alanc> that's up to whatever protocol you're running on top of it[01:08:22] <alanc> all the socket says is that you get data from one point to another, "requests" are higher level protocol constructs[01:08:32] <sergefonville> so the fifo has nothing to do with the socket[01:08:35] <alanc> it's not an RPC mechanism, just a communications channel[01:09:29] <alanc> it's fifo in that the first data written by the sender is the first data read by the reciever[01:09:49] <alanc> (excluding the special case of UDP)[01:09:58] <sergefonville> ah, that's what you meant, makes sense[01:11:08] <alanc> each client connection to a server is represented as a seperate file descriptor - it's up to the server to decide which order to read/process data on each fd[01:11:27] <sergefonville> that explains a lot[01:11:53] <alanc> it can have a thread per connection, or multiplex through a select/poll loop, etc.[01:16:49] <Shadow__X> anyone here have experience with a lsi 9211[01:21:01] *** freedomrun has quit IRC[01:21:43] <sergefonville> thanks for explaining[01:21:48] <sergefonville> I'm off[01:21:56] *** sergefonville has left #openindiana[01:24:08] *** hjf_ has joined #openindiana[01:25:45] *** hjf has quit IRC[01:31:04] *** hjf_ is now known as hjf[01:33:37] <DanaG> hmm, 2.5MB/second copying fron windows ntfs to zfs NAS.[01:33:44] <DanaG> Is that low for gigabit ethernet?[01:33:51] <Shadow__X> 2.5MB?[01:33:55] <DanaG> how do I tell what's the bottleneck?[01:34:07] <DanaG> Oh, 60% cpu usage on the computer... that's what.[01:34:52] <DanaG> Looks like a non-optimal utility.[01:35:21] <DanaG> ah, 38MB/sec now. That's better.[01:35:24] <hjf> DanaG: I get around 80MB/s[01:36:19] <Shadow__X> yeah on my zfs build using zfs-fuse on linux with a not so great highpoint card i was getting around 50 write from my raid5 md array[01:36:32] <Shadow__X> so i would expect to get higher than 50 or so easily[01:36:53] <hjf> Well I get 240MB/s local writes with just the embedded SATA controller[01:36:55] <Shadow__X> i think past somehwere around 60-70 you probably need a better nic or atleast in my situation i do[01:37:17] <hjf> my 80MB/s NIC was the nvidia nforce gigabit ethernet, mobo onboard[01:37:19] <Shadow__X> hjf: how do you test local writes?[01:37:47] <Shadow__X> i have one nforce asus p5n-d and my other machine uses a realtek nic i think[01:37:55] <hjf> Shadow__X: dd from /dev/zero[01:38:07] <Shadow__X> did you time it?[01:38:22] <hjf> oh no 240MB/s were reads. writes were a little less but well over 160 I'm pretty sure[01:38:35] <hjf> Shadow__X: if you use GNU dd it times for you[01:38:44] <hjf> (use gdd instead of dd)[01:39:17] <Shadow__X> ah ok i use hdparm normally[01:39:30] <hjf> yeah on linux[01:39:39] <hjf> there's no hdparm -Tt in solaris[01:39:48] <Shadow__X> i gotcha[01:40:14] *** InTheWings has quit IRC[01:40:16] <hjf> dd /dev/zero is not the ideal tool to benchmark either[01:40:18] <Shadow__X> so something like dd if=/dev/zero of=/mnt/folder/file what do you use as your byte size[01:40:29] <hjf> bs=1M :P[01:41:41] <Shadow__X> then you just cancel it?[01:41:53] <hjf> yes ctrl-C it after 10 secs or so[01:42:23] <hjf> Bonnie++ would probably be a more accurate tool[01:42:42] <Shadow__X> 250MB/sec on my md raid[01:42:57] <Shadow__X> on my old machine that runs openindiana i am sure its really slow[01:42:58] <hjf> Shadow__X: how many disks? sata?[01:43:34] <Shadow__X> well that machine is 5 disks in raid 5, sata 2 using mdadm on ubuntu[01:44:10] <hjf> good. remember that ZFS is not about raw performance though[01:44:15] <hjf> it's about reliability[01:44:33] <hjf> that's why you have checksumo everywhere[01:44:35] <Shadow__X> yup the 2 machines i am running zfs on are for backups[01:44:37] <hjf> checksums[01:44:53] <hjf> same[01:45:16] <Shadow__X> the current box i am bringing up will hold backups of multiple machines so i care about my data being safe[01:45:43] <hjf> Shadow__X: this is what I'm currently using: http://www.mswhs.com/wp-content/uploads/2010/10/HP-Microserver.jpg[01:46:24] <Shadow__X> oh cool i think i saw one of those on newegg wasnt it like 400 or 500 bucks? or was that another one[01:46:41] <hjf> yes[01:47:31] <Shadow__X> sweet, does each drive have led notifications? so for example if a drive is failing would a led switch to red or something[01:47:40] <hjf> nope :P[01:47:51] <hjf> just write down the drive serial numbers in each tray[01:48:23] <Shadow__X> and then use openindiana to send an email about a drive?[01:48:42] <hjf> yes[01:49:32] <Shadow__X> how bad was it to run openindiana on it? and how are the speeds[01:49:57] *** jaimef has quit IRC[01:50:09] <hjf> Shadow__X: I don't use OI :P I use S11 express. just cause I'm lazy and never got around to upgrade to OI[01:50:16] <hjf> it Just Worked out of the box[01:51:27] <Shadow__X> but openindiana should work much the same way right? How are the network speeds and uptime?[01:52:49] <hjf> network speeds were OK with the shitty onboard NIC (bge) but it doesn't have jumbo frames[01:53:11] <Shadow__X> nice and even under high loads its been good? hmm i have seen over 50MB/sec without jumbo frames[01:54:29] <hjf> Shadow__X: not sure, I'm just a single user lol[01:54:54] <Shadow__X> oh ok well what were the network speeds with the integrated one under 50>?[01:55:37] <Shadow__X> also does the boot drive get installed in one of the 4 hotswaps or is there an internal spot for it[01:56:22] <hjf> Shadow__X: I don't have a Gigabit desktop anymore (downgraded to a laptop) so I can't test right now but I'm pretty sure it reached 50MB easily[01:56:54] <hjf> it's not hotswappable, according to the manual. But i hotswapped drives once on one of those running linux and it worked just fine[01:57:16] <hjf> there are 2 extra SATA ports: 1 is eSATA (back) and 1 "ODD" (optical disk drive)[01:58:00] <hjf> these last 2 run in IDE emulation and are stupid slow. some russian dude hacked the BIOS and you can change them to AHCI. I did that and it works great[01:59:38] <hjf> i connected the boot drive to the ODD connector, and just throwed it into the DVD drive hole.[02:00:44] <Shadow__X> oh ok does the esata port support port multipliers?[02:01:47] <Shadow__X> btw thanks for the info hjf, its really appreciated[02:01:54] <hjf> Shadow__X: I don't think so. this server is as baseline as it gets. also the AHCI driver from solaries doesn't support PMPs[02:02:08] <Shadow__X> oh ok[02:02:12] <hjf> from S11X[02:02:17] <hjf> not sure about OI ...[02:02:35] <Shadow__X> i gotcha is the ram upgradable?[02:02:50] <hjf> yes and you better do it cause it has just 1GB[02:03:05] <Shadow__X> :) yeah. one slot or two?[02:03:05] <hjf> it has 2 ECC DDR3 dimms[02:03:15] <Shadow__X> oh cool thats a huge plus[02:03:17] <hjf> it supports ECC and Non-ECC ram[02:03:33] <hjf> i got an extra 4GB from it. Kingston generic[02:03:45] <Shadow__X> whats the suggested value 1gb for every tb?[02:03:49] <hjf> KVR1333D3E5[02:04:10] <hjf> no idea but most guys here laugh at the 8GB the microserver supports[02:04:21] <hjf> 24G seems to be commonplace around here[02:04:42] <Shadow__X> but these are production servers right?[02:04:54] <Shadow__X> by these i mean the ones that most people around here run[02:05:00] <Shadow__X> clearly not the hp[02:05:04] <hjf> yeah although you have some crazy types that have 24G at home[02:05:30] <hjf> how many users will you have accessing that server?[02:07:13] <miine> I think it's more questionable if the hp microserver is a smart choice at all: it's motherboard is custom form factor...[02:07:23] <Shadow__X> at this point i would be thinking of having it as a dedicated backup server so really just one but in the even the main server goes down a total of maybe 5 concurent users[02:07:54] <Shadow__X> miine: even price wise?[02:08:33] <hjf> miine: it's $300 and it consumes 10W or so[02:08:42] <miine> at the moment the Dell T110 is sold at really low...[02:09:05] <miine> hjf: 10W is really great, but the Dell consumes 40-60W if idle...[02:10:09] <hjf> let me see... apcupsd reports 11% load. it's an APC BR1000I (1000W/600VA) and there's also the router, modem, switch, another modem, and IP Camera[02:10:57] <miine> but given the form factor of the case and prices of mini itx boards and the chembro cases the microserver looks good...[02:11:14] <miine> the only thing I don't like is that the mobo isn't standard...[02:11:33] <hjf> miine: well you can buy 2 and keep one in storage :P[02:12:28] <hjf> and it runs cooooool[02:12:36] <hjf> it has a single, huge fan in the back[02:13:10] <hjf> I wouldn't mind having the server in my bedroom[02:13:56] <hjf> even with all the disks running (WD Greens) I can barely hear it[02:14:05] <Shadow__X> hmm if dell lets you buy the t110 base without a hard drive it would be 200 bucks[02:14:08] <miine> hjf: does it have an internal usb socket for a stick?[02:14:13] <mnaser> i'm setting up a server in a couple of days[02:14:16] <hjf> miine: yes[02:14:40] <mnaser> 42 1TB drives, 2 mirrored 32GB ZILs, 2x 160GB L2ARC, 48GB RAM on head node[02:14:43] <mnaser> + infiniband[02:15:07] <Shadow__X> mnaser: please tell me its for home use :)[02:15:21] <hjf> miine: http://www.mswhs.com/wp-content/uploads/2010/10/HP-Microserver.jpg it's right there at the front left of the PCB the silver square next to the brown...thing[02:15:25] <miine> Shadow__X: the cool thing about dell is: it comes with all cables / trays for up to 4 drives, you can opt for an i3 and 3 years next-day on-site is cheap...[02:15:40] <mnaser> haha, overkill, 21TB usable space (will raid-10)[02:15:50] <mnaser> vm hosting with xenserver[02:15:59] <miine> hjf: thanks...[02:16:03] <Shadow__X> miine: yeah[02:17:12] <hjf> miine: ah yes the microserver has a really shitty CPU[02:17:23] <miine> and as the dvd-rom is also sata, put in a 5.25" hd carrier and a fifth hd. so mirrored rpool and 3 drives RAID-Z1 ...[02:17:29] <hjf> a low power athlon 64 (athlon II neo,AMD's answer to the intel atom)[02:17:59] <miine> hjf: which should be way faster then an atom 510/330 or isn't it ?[02:18:19] <miine> hjf: not to forget ECC ram...[02:19:00] <hjf> yes, cause it's the same thing as an athlon II. it's just a low power version with all the features[02:19:30] <hjf> I use it as my home server. I'm a photographer and I back up my files there. so i keep a copy on my laptop and other on the server[02:19:36] <Shadow__X> at one point i was trying to compare the validitiy of building my own cheaper machine but with the inclusion of ecc i am unsure i could get something together in the same price range as the hp micro[02:20:16] <hjf> Shadow__X: most athlon64/athlon II/phenom/etc mobos support ECC. you just need to plug ECC memory to them[02:20:55] <miine> Shadow__X: the microserver's price is fine. removable hd's etc. : the case with power supply would normaly cost you half of microserver's price alone.[02:20:57] <hjf> it's intel who thinks you shouldn't get ECC with the Core series. yeah, cause the core i7 is so awesome that people would be using it for servers instead of Xeons[02:21:22] <miine> Not to forget that decent atom mobo's with > 4x sata cost about 180 bucks too...[02:21:33] <Shadow__X> i did not know that about most amd cpu's[02:22:11] <hjf> Shadow__X: intel has a shitload of versions: with/without video, with/without AES, with/without Intel-VT[02:22:25] <hjf> amd just has different speeds. all their athlons are the same[02:22:49] <Shadow__X> yeah i know about intel at this point its hard to keep track[02:22:49] <miine> but even using more standard (cheap) components, the last time I checked I come up at roughly the same price. And then I say the Dell offer ;-)[02:22:58] <miine> say -> saw ...[02:23:17] <Shadow__X> hmm but with building your own you would be able to fit more drives though right?[02:23:44] <hjf> Shadow__X: not for that price[02:23:57] <hjf> most decent mobos give you 6x SATA ports[02:24:15] <miine> Shadow__X: of course. depends on the case and what type of additional controller you put in...[02:24:20] <Shadow__X> i tried a proof of concept running mac zfs on a drobo :) so far when enabling and disabling compression i get kernel panics. I blame drobo though[02:24:31] <hjf> if you want more you need expensive SATA controllers. SiL controllers don't work with osol/OI. it hangs.[02:25:35] <miine> Shadow__X: mac zfs from google? maybe ten complement's will work better on that regard...[02:26:28] <hjf> Shadow__X: if using ZFS, use solaris/OI. don't use linux, freebsd, or anything else[02:26:33] <miine> hjf: not true. I'm having a SiL 3112 (2 ports) in my atom 330 here. did run osol 09/06 fine, and runs oi148. but as it's pci only, its slooow...[02:27:28] <hjf> miine: I tried with SiL 3124 (pci-express) and 3132 (pci) and never got it to work decently[02:27:31] <miine> hjf: the problem might be the firmware on those sil cards. mine is a Mac version, so the firmware doesn't get loaded at all :-)[02:28:06] <hjf> miine: nope, I tried different firmware versions (with/without "hardware" RAID)[02:28:09] <miine> hjf: did you disable everything not needed in the pc-bios? (Parallel, serial ports etc..)[02:28:27] <Shadow__X> hjf: i understand not recommending linux yet as its a fuse driver thats still under development(i tried it and it was stable just slow) but why not freebsd[02:28:48] <Shadow__X> miine: could you reflash it[02:28:58] <miine> hjf: I think that might have been interrupt issues. but anyway: I would always prefer onboard sata as on the controller side otherwise it might getting expensive..[02:29:01] <hjf> miine: yes. i spent weeks trying to debug it. the support here was "just get a better controller"[02:29:56] <hjf> Shadow__X: there is a kernel vesion (zfsonlinux.org) but it's much slower[02:30:21] *** Botanic_ has joined #openindiana[02:30:25] <hjf> Shadow__X: freeBSD is ok (i have one running) but... freebsd is complicated (power management, driver support, etc)[02:30:31] *** Botanic has quit IRC[02:30:44] <Shadow__X> really? compared to oi[02:31:46] <miine> is anyone using puppet here? how do zones get handled - can a simply make a node require a zone from the global zone'z node?[02:31:47] *** held has quit IRC[02:32:05] *** trn has quit IRC[02:32:29] *** held has joined #openindiana[02:32:40] <hjf> Shadow__X: well, I got 16% load on my UPS when idle (linux) vs 11% (solaris 11 express)[02:33:01] <Shadow__X> oh wow nice just by using defaults?[02:33:04] <hjf> and linux had the silly tickless kernel and all power optimizations you could think of. solaris was just default[02:33:43] *** LeftWing has quit IRC[02:33:46] <Shadow__X> oh ok i gotcha[02:34:00] <Shadow__X> is there wol for oi?[02:34:34] *** LeftWing has joined #openindiana[02:34:42] <hjf> Shadow__X: depends on the NIC. i never got it to work with nge[02:34:46] *** Botanic_ has quit IRC[02:35:01] *** trn has joined #openindiana[02:35:34] <Shadow__X> which is that[02:35:43] *** DrLou has joined #openindiana[02:35:43] *** ChanServ sets mode: +o DrLou[02:35:46] <hjf> nvidia gigabit ethernet[02:35:50] <hjf> Shadow__X: man nge ;)[02:35:50] <Shadow__X> oh ok[02:36:04] <Shadow__X> i dont have a oi machine up right now[02:47:16] *** anAnunaki has quit IRC[02:48:08] *** trn has quit IRC[02:57:58] *** trn has joined #openindiana[02:58:52] *** McBofh has quit IRC[03:03:24] *** DrLou has quit IRC[03:06:55] *** McBofh has joined #openindiana[03:11:20] *** master_of_master has quit IRC[03:13:34] *** master_of_master has joined #openindiana[03:18:02] *** TBCOOL has quit IRC[03:31:45] *** ThomasB2k has quit IRC[03:32:39] *** ThomasB2k has joined #openindiana[03:39:49] <DanaG> hmm, is there an equivalent of cpufreq-info on OI?[03:44:31] <DanaG> hmm, now I have one computer copying stuff via "richcopy" over samba, and another copying via Unison. TOtal is 40 megabytes per second.[03:44:50] <DanaG> This seems to be cpu-limited by the clients.[03:45:18] <DanaG> Oh, and is there any conceivable reason one would want to go from an Intel card back to the Broadcom 5723? Aside from the activity LED, that is.[03:46:32] <Shadow__X> arent intel cards regarded to be some of the better cards[03:46:53] <DanaG> yeah.[03:59:18] <DanaG> Now I just need to find a DAAP server package for OI.[04:05:24] *** TBCOOL has joined #openindiana[04:08:03] *** kart_ has joined #openindiana[04:10:49] *** dijenerate has quit IRC[04:29:03] <DanaG> okay, anyone know of a DAAP server that'll work on OI, or do I really need to have a Linux VM within vbox?[04:29:07] <blues> so for an inital stab at OI + nappit, i've got 4x 2TB drives to work with. I want to offer up a iSCSI connection to ESXi host for vms then share out rest of the space via NFS shares... What would you suggest as my pool config?[04:29:08] <DanaG> That seems inefficient.[04:29:23] <DanaG> my thing, not yours, inefficient.[04:33:41] *** sivanov has quit IRC[04:34:11] *** sivanov has joined #openindiana[04:38:23] *** POloser has joined #openindiana[04:38:54] <Triskelios> SFEmt-daapd?[04:42:26] *** redgone has joined #openindiana[04:46:18] <blues> DanaG: isn't one of the points of Vm being able to specialize a bit so that you have less overlap? I dunno what the package situation is like w/ BSD but you know there's a large amount of suitable software on the linux side, where's the harm in throwing out a virtualized instance just for your media sharing ?[04:51:34] <sivanov> ok, now at home i get 89MB/s write, 112MB/s read from cifs share via iozone[05:01:46] <DanaG> argh, my microserver is making an obnoxious high-pitched tone.[05:02:32] <DanaG> Something like 9 kHz[05:03:49] *** jaimef has joined #openindiana[05:04:39] <jaimef> is it possible to have a OS 2009 BE preserved after upgrading to say 151?[05:07:05] *** gore has quit IRC[05:07:26] *** gore has joined #openindiana[05:08:05] <geoff_> if you image-update, then the old BE's stay around, until you manually delete them.[05:10:01] <jaimef> k. audiohd regression on mcp79 is the only reason I would even both with OS[05:15:41] *** miine has quit IRC[05:16:41] *** sivanov is now known as sivanov_[05:17:24] <sivanov_> and if you upgrade zpool version you won't be able to access old be[05:19:32] <sivanov_> i get like 14000 interrupts on bge0 while windows initiator runs iozone @ my oi target[05:19:47] <sivanov_> isnt it a bit to high, is it?[05:33:55] *** DontKnwMuch has quit IRC[05:43:16] *** skeeziks has quit IRC[05:52:51] *** Naresh has quit IRC[05:58:25] *** redgone has quit IRC[06:16:37] *** tsukasa has quit IRC[06:17:37] <dkeav> sivanov_: oi148?[06:18:54] <sivanov_> yes[06:18:58] <dkeav> there are patches for broadcom chipset, do an image-update[06:19:36] <sivanov_> i did, before it i was getting 100K irqs and 50% cpu usage[06:19:53] <dkeav> well 14k isn't so bad then hehe[06:20:52] <sivanov_> iozone @CIFS share http://www.pastie.org/2194959[06:21:31] <sivanov_> iozone @iscsi disk: http://www.pastie.org/2194966[06:21:56] <sivanov_> not bad, but i cant nearly max FC on that server[06:22:38] <sivanov_> getting 100MB/s lower write and 200MB/s lower read than to local disk[06:23:22] <blues> i'm ready to setup my first iscsi experiement with OI/napp-it. ANy good tutorials online for this/[06:23:23] <blues> ?[06:25:42] *** ThomasB2k has quit IRC[06:26:54] <sivanov_> so, you flashed your controler, have you?[06:27:20] <blues> yeah.. all it took was (apparently) frying one motherboard and mixing and matching a shelf's worth of parts[06:27:51] <blues> i still dunno what is up with that damn motherboard... i can't see how i fried it.. unless there's some big no no about trying to use integrated video and putting the controller card in the x16 slot at the same time[06:28:39] <Shadow__X> blues: i had a motherboard that was working for over 3 years die after all i did was plug in a pcie card in[06:28:50] <Shadow__X> a card that i had used in the computer before[06:29:07] <DanaG> Cool, I found an mt-daapd that works.[06:29:10] *** tsukasa has joined #openindiana[06:29:17] <blues> yeah i mean... SOMETHIng happened.. i've been playing with these things for long enough to just accept that shit happens.[06:29:38] <blues> Waiting on RMA on that motherobard will just slow me up[06:29:48] <DanaG> I have a motherboard that survived a PSU death, but seems to have gotten +5V fused to +5Vsb.[06:30:04] <sivanov_> right now im looking @ ASUS DSBF-D16/SAS motherboard on ebay[06:30:29] <DanaG> So, all I need now is deluge (or perhaps transmission will do).[06:30:54] <sivanov_> lol, i fried a PSU connected to motherboard by not setting 110-240V switch to right position[06:31:07] <sivanov_> the mobo surprisengly survided[06:33:10] <sivanov_> i want that motherboard, however i have predjustice towards ASUS[06:33:49] <sivanov_> i have like 4 asus motherboards dead on my hands[06:33:56] <sivanov_> in 2 years[06:34:29] <blues> Well, in any case, i'm finally ready to start getting dirty with this thing. I've got 4 disks installed..i'm thinking i should mirror 2 of them for my VM storage.. any problems with that?[06:35:03] <sivanov_> why not mirror 4?[06:35:14] <sivanov_> you'll have more aggregate IOPS[06:35:39] <blues> What is the real-raid equiv of mirroring 4 ?[06:35:45] <blues> 10?[06:37:10] <sivanov_> yes[06:37:53] <sivanov_> if you add another raid group later you will need some copying&removing of files[06:38:17] <sivanov_> to balance the data accross new&old vdevs[06:39:09] <blues> for the iops, i dunno that i need to maximize with this home-lab setup[06:39:42] <blues> but i can't really do much with the other two disks eitehr, until i add the rest of my storage[06:40:55] <sivanov_> in case you have zvols exported by comstar it may be inconvinient as you will have to send&receive the zvols and then to re-create the luns with[06:41:19] <sivanov_> and vmware may go pricky about the disk IDs[06:42:58] <sivanov_> but again to that mobo: it has lsi 1086 pci-x sas controller, apparently it works in raid more[06:43:00] <sivanov_> *mode[06:43:18] <sivanov_> how i flash it to IT mode??[07:02:15] *** Naresh has joined #openindiana[07:05:09] *** dijenerate has joined #openindiana[07:09:46] <madwizard> Coffee[07:12:42] <DanaG> say, where do I get "libao" headers?[07:15:51] <DanaG> I need it to compile shairport[07:16:24] <bdha> Triskelios: Oh, right. You're Albert Lee. Saw the LinkedIn req and was confused. I work with an Albert Lee, too.[07:17:38] <DanaG> It also needs pkg-config[07:17:38] <alanc> I'm sure I've rejected linkedin requests from people who I only know by IRC nick and have no clue who the hell they are by real name[07:17:51] <alanc> pkg-config is part of the gnome development package[07:19:07] *** jaimef` has joined #openindiana[07:19:10] <jaimef`> j #illumos-dev[07:19:28] <alanc> /[07:19:42] <DanaG> Which actual package?[07:19:59] <alanc> the one that "pkg search pkg-config" tells you[07:20:19] <DanaG> ah, missed that it tells you which package.[07:20:20] <alanc> I don't have them all memorized, so would have to go use pkg search myself, so you might as well do it[07:20:22] <bdha> alanc: I just don't log into linkedin. :)[07:20:37] <bdha> Unless someone badgers me.[07:21:21] *** McBofh has quit IRC[07:21:31] *** Botanic has joined #openindiana[07:21:38] <alanc> well, if you worked at a company where half your coworkers are leaving, you might go a bit more often to find out where they all end up[07:21:48] <bdha> Yeah, fair enough. :)[07:22:22] <DanaG> okay, no package has ao.h[07:22:33] *** forquare has joined #openindiana[07:29:43] <DanaG> So, where do I get libao devel headers?[07:31:30] *** jaimef` has quit IRC[07:31:30] *** freedomrun has joined #openindiana[07:32:45] <DanaG> assertion failed: RAND_MAX > 0x10000[07:32:51] *** gnomad has joined #openindiana[07:32:55] <bdha> ...build it yourself?[07:33:48] *** forquare has quit IRC[07:34:00] <DanaG> Actually, I found a package in the form of .tgz, and extracted it to root.[07:34:23] <alanc> since the headers would come in the package with the library, lack of headers is likely not the big problem[07:35:13] <DanaG> pkgsearch-repo.uk.openindiana.org/2011Q1/devel[07:35:41] <DanaG> hmm, no libwww-perl, either.[07:36:39] <bdha> ah, yeah. The IPP repo. Good call.[07:36:42] <bdha> (pkgsrc -> IPS)[07:36:55] <DanaG> Can I just plain add that repo?[07:37:04] <DanaG> And what does ipp stand for?[07:37:11] <bdha> illumos pkgsrc project.[07:37:28] <bdha> https://www.illumos.org/projects/worsoe[07:40:34] <DanaG> cool, so now all I need is whatever has HTTP/Request.pm[07:40:42] <DanaG> Well, that, and I need to figure out what's up with RAND_MAX[07:41:38] <alanc> it's the universe's way of telling you to use a much less crappy random number generator than rand(3C)[07:42:04] *** McBofh has joined #openindiana[07:43:24] <alanc> because you can't change RAND_MAX, just use a different RNG with a higher MAX and better algorithm[07:44:01] <DanaG> It's somebody else's code, so I don't know what it's doing.[07:46:17] <alanc> if they're using RAND_MAX, then the author doesn't know what they're doing either[07:46:30] <DanaG> Anyway, the HTTP/Request.pm is a bigger blocker.[07:48:26] <alanc> CPAN should work fine for finding & installing perl modules[07:50:45] <DanaG> Here's the sound my microserver makes: https://www.dropbox.com/s/nvxzf8f2kjn9nea/aaaugh.wav[07:50:50] <DanaG> Notice: it's named that for a reason.[07:50:57] <DanaG> er, wrong channel.[07:52:30] *** jaimef` has joined #openindiana[07:52:58] *** freedomrun has quit IRC[07:53:24] *** jaimef` is now known as akkad[07:56:29] *** Crypticfortune has quit IRC[08:10:14] <DanaG> Anyway, I give up on shairport for now.[08:16:27] *** hajma_ has quit IRC[08:24:45] *** gea__ has joined #openindiana[08:25:02] *** akkad has quit IRC[08:27:21] <sivanov_> how can i do something like running tests on local zvol, partitioned with gpt?[08:27:43] <sivanov_> where to look for gpt slices?[08:35:49] *** gea__ has quit IRC[08:44:33] <DanaG> So tomorrow, I'll set up an smf file for that mt-daapd, as well as for Transmission daemon.[08:44:33] *** joes_ has joined #openindiana[08:45:34] *** dekar__ has quit IRC[08:46:33] *** merzo has joined #openindiana[08:50:07] *** dekar__ has joined #openindiana[08:56:43] *** anikin has joined #openindiana[08:57:33] *** miine has joined #openindiana[08:57:34] <sivanov_> i need a dlna server for OI which works with yamaha 660NX(?) receiver[08:58:40] <sivanov_> i wonder, if fuppes will work: iirc, it needed pcap[09:06:13] *** |AbsyntH| has joined #openindiana[09:15:21] <gea_> @blues: what sort of tutorial are you looking for?[09:20:51] *** Micr0mega has joined #openindiana[09:25:28] *** hajma has joined #openindiana[09:28:44] *** gber has joined #openindiana[09:28:57] *** raichoo has joined #openindiana[09:29:44] <raichoo> hi folks[09:31:06] *** freetown2 has joined #openindiana[09:31:48] <freetown2> hi, is there anything similar to iptables -m state --state ESTABLISHED,RELATED in ipfilter?[09:41:28] *** bens1 has joined #openindiana[09:42:56] *** gea_ has quit IRC[09:45:17] *** gea_ has joined #openindiana[09:47:38] *** McBofh has quit IRC[09:56:57] *** lblume1 is now known as lblume[10:03:51] *** DanaG has quit IRC[10:04:59] *** freedomrun has joined #openindiana[10:08:06] *** lkocman has joined #openindiana[10:24:06] *** gervystar has joined #openindiana[10:26:34] *** kart_ has quit IRC[10:27:42] *** kart_ has joined #openindiana[10:33:26] *** ThomasB2k has joined #openindiana[10:47:33] *** sivanov_ has quit IRC[10:48:03] *** sivanov_ has joined #openindiana[10:49:56] *** bgeek has quit IRC[10:51:03] *** ArnaudEC_ has quit IRC[10:51:17] *** ArnaudEC has joined #openindiana[10:51:33] *** Whoopsie has joined #openindiana[10:51:33] *** ChanServ sets mode: +v Whoopsie[10:51:34] *** Andrew1 has quit IRC[10:51:49] *** movement has quit IRC[10:52:51] *** Andrew1 has joined #openindiana[10:58:03] *** reddi has joined #openindiana[11:00:25] *** davenz has joined #openindiana[11:05:34] *** kart_ has quit IRC[11:07:33] *** movement has joined #openindiana[11:17:36] *** gervystar has quit IRC[11:21:51] *** dekar__ has quit IRC[11:27:55] *** descipher_ has joined #openindiana[11:28:24] *** Naresh` has joined #openindiana[11:30:26] *** taemun_ has joined #openindiana[11:30:52] <sivanov_> o.O can i update a S11E package with illumos package?[11:31:26] <sivanov_> looks like S11E iscsi target has a bug fixed in illumos[11:31:36] *** jkj_ has joined #openindiana[11:32:14] *** iLukeBer1y has joined #openindiana[11:32:46] *** mikaeld_ has joined #openindiana[11:35:09] *** man_u_ has joined #openindiana[11:36:08] *** miine_ has joined #openindiana[11:36:26] *** |AbsyntH| has quit IRC[11:36:26] *** miine has quit IRC[11:36:26] *** anikin has quit IRC[11:36:26] *** Naresh has quit IRC[11:36:27] *** freetown2 has quit IRC[11:36:28] *** descipher has quit IRC[11:36:29] *** robinbowes has quit IRC[11:36:31] *** iLukeBerry has quit IRC[11:36:31] *** taemun has quit IRC[11:36:32] *** mikaeld has quit IRC[11:36:32] *** papertigers has quit IRC[11:36:33] *** easyE has quit IRC[11:36:34] *** jkj has quit IRC[11:36:34] *** Warod has quit IRC[11:36:34] *** man_u has quit IRC[11:36:35] *** miine_ is now known as miine[11:36:35] *** taemun_ is now known as taemun[11:36:35] *** man_u_ is now known as man_u[11:36:35] *** Naresh` is now known as Naresh[11:36:36] *** McBofh has joined #openindiana[11:36:36] *** kinesis_ has joined #openindiana[11:36:43] *** xmikus01 has joined #openindiana[11:38:58] *** Naresh has quit IRC[11:38:58] *** Naresh has joined #openindiana[11:41:00] *** Warod has joined #openindiana[11:42:28] *** papertigers has joined #openindiana[11:43:01] *** davenz has left #openindiana[11:43:33] *** |AbsyntH| has joined #openindiana[11:43:38] *** robinbowes has joined #openindiana[11:44:00] *** held has quit IRC[11:44:59] *** freetown2 has joined #openindiana[11:47:23] *** bgeek has joined #openindiana[11:50:29] *** flyz has joined #openindiana[12:03:23] *** held has joined #openindiana[12:06:28] *** gea__ has joined #openindiana[12:10:58] *** DeanoC has joined #openindiana[12:10:58] *** ChanServ sets mode: +o DeanoC[12:17:04] *** gea__ has quit IRC[12:23:10] *** |AbsyntH| has quit IRC[12:49:20] *** InTheWings has joined #openindiana[12:50:37] *** adama has quit IRC[13:08:10] *** kart_ has joined #openindiana[13:10:52] *** mikaeld_ has quit IRC[13:11:15] *** mikaeld has joined #openindiana[13:15:25] *** Killsudo has quit IRC[13:16:45] *** Killsudo has joined #openindiana[13:18:46] *** Micr0mega has quit IRC[13:30:23] *** Micr0mega has joined #openindiana[13:49:18] *** DrLou has joined #openindiana[13:49:18] *** ChanServ sets mode: +o DrLou[14:10:58] *** dekar has joined #openindiana[14:26:37] *** POloser has left #openindiana[14:29:18] *** Whoopsie has quit IRC[14:38:54] <kinesis_> I recently complained about the idmap/smb error.[14:39:06] <kinesis_> now it happens again[14:39:21] <kinesis_> i cant stop smb/server service[14:39:27] <kinesis_> until reboot.[14:39:54] <kinesis_> Jul 11 16:11:18 box idmapd[3753]: [ID 280452 daemon.error] Error: smb_lookup_sid failed.[14:39:54] <kinesis_> Jul 11 16:11:18 box idmapd[3753]: [ID 455671 daemon.error] Check SMB service (svc:/network/smb/server).[14:39:54] <kinesis_> Jul 11 16:11:18 box idmapd[3753]: [ID 174421 daemon.error] Check connectivity to Active Directory.[14:40:25] <kinesis_> anyone have any thoughts?[14:43:18] <kinesis_> no ideas?[14:43:58] <sivanov_> check time[14:44:50] <sivanov_> on opensolaris and AD controllers[14:45:39] <viridari> kinesis_: how are you trying to stop it?[14:47:37] <kinesis_> trying svcadm disable smb/server[14:48:08] <kinesis_> time synchronized with DC by cron.[14:48:45] <kinesis_> first time i get this error 2 weeks ago.[14:49:17] <kinesis_> i have 2 DCs and 2 IO_148 servers.[14:49:29] <kinesis_> only one have this error ([14:49:41] *** radsy has joined #openindiana[14:50:02] <sivanov_> can i do something with GPT disk from another system in OI?[14:50:41] <sivanov_> it looks like format does not understand the partition table...[15:00:24] *** hajma has quit IRC[15:00:52] *** hajma has joined #openindiana[15:04:28] <kinesis_> could it be memory error?[15:07:17] <tsoome> someone have asked about that smb_lookup_sid here already[15:08:04] <kinesis_> i think it was me[15:08:10] <tsoome> kinesis_: use ntp, not just cron[15:08:16] <kinesis_> last week[15:08:32] <tsoome> ah, might be. well, how it did resolve then?[15:08:35] <kinesis_> i use ntpdate in cron. its now right?[15:08:45] <tsoome> well, it will sync, sure[15:09:27] <kinesis_> i cant resolve it. to start service working i can only restart server.[15:09:34] *** AlasAway is now known as Alasdairrr[15:09:50] <tsoome> but with time syncing its better to have ntpd up and syncing, because ntpdate can still make time jumps (if by some reason the clocks are drifting). if you have good clocks, then that shouldnt happen:)[15:10:35] <tsoome> I also have no idea how large window is allowed in kerberos...[15:10:44] <tsoome> time window, i mean[15:10:57] <kinesis_> i dont think its time problem. time was syncked this night.[15:11:20] <kinesis_> kerberos have 10 minutes window i think.[15:11:34] *** lkocman has quit IRC[15:11:46] <tsoome> I guess that too, or else you would have more time related messages i guess[15:12:14] <kinesis_> it happens suddenly.[15:12:36] <kinesis_> for a half-year server worked perfectly.[15:13:30] <tsoome> so, then its the question, what did change - maybe some updates were applied on AD server?[15:14:25] <kinesis_> no, updates disabled. and second server still working fine.[15:14:44] <kinesis_> same configuration and OI_148[15:15:10] <tsoome> also you can test that kerberos connection with kinit/klist[15:15:37] <tsoome> to see if you can login to kerberos and get ticket for your user...[15:15:57] <sivanov_> weird, cant get speed above 160MB/s on read over FC4 :([15:16:16] <sivanov_> basicly same speed as FC2[15:16:19] <kinesis_> 6692213 smbd crashes in smb_lookup_sid[15:16:20] *** |AbsyntH| has joined #openindiana[15:16:36] <tsoome> ah, so its known bug then...[15:16:45] <kinesis_> its from 3 years old changeset[15:17:25] <kinesis_> now I can not find anything like it.[15:18:15] <kinesis_> may be i need to update system?[15:18:44] <tsoome> only thing you can update to is 151 dev build....[15:19:23] <tsoome> or just wait till 151 will be released...[15:20:04] <sivanov_> is it possible to update a package, say iscsi/target on S11E from oi_151??[15:20:16] <kinesis_> my users will eat me before )[15:21:25] <tsoome> sivanov_: dont think so. but if you have S11, you can update from support repo...[15:21:41] <sivanov_> haha[15:21:43] <sivanov_> no ty :D[15:22:36] <sivanov_> iscsi target S11E+initiatror Centos 5.2=unsuccess[15:23:09] <sivanov_> the bug is apparently fixed in illumos but not in S11E[15:23:15] <sivanov_> btw[15:23:31] <sivanov_> cifs__speed: http://www.pastie.org/2194959[15:23:31] <sivanov_> iscsi_speed: http://www.pastie.org/2194966[15:23:32] <sivanov_> FC2___speed: http://www.pastie.org/2195348[15:23:32] <tsoome> ?[15:24:00] <tsoome> if you dont have support, how can you tell its not fixed?[15:24:05] *** buffyg has quit IRC[15:24:51] <sivanov_> i mean, S11E has this bug[15:25:02] <sivanov_> apparently[15:26:16] *** lkocman has joined #openindiana[15:28:00] <sivanov_> FC4 speed: http://www.pastie.org/2196666[15:28:55] <sivanov_> i think that something is wrong with FC4 read speed that FC2 gives actually better read performance[15:29:07] <sivanov_> also FC2 is NTFS under win 7[15:29:31] <tsoome> what is your target HBA?[15:29:49] <sivanov_> qla2460[15:30:39] <sivanov_> also switch shows some errors on 2GB port, matching speeds meh[15:30:58] <sivanov_> no errors on initiator 4GB port[15:31:49] *** mikw has joined #openindiana[15:32:19] <tsoome> fcinfo hba-port -l does list errors as well?[15:33:18] <tsoome> its using qlt for target driver, right?[15:33:31] <sivanov_> yes, and no errors[15:33:34] *** syoyo_ has joined #openindiana[15:34:06] <sivanov_> will run read test connecting to target via iscsi loopback interface[15:34:12] <tsoome> hm, sru 8 readme does not mention qlt[15:34:16] *** andygraybeal has joined #openindiana[15:35:19] <sivanov_> just dont want to try emulex 12k which i also have (surplus from work hehe)[15:35:57] <sivanov_> emlx used to give me fct buffer errors, so i got rid of it[15:38:17] <tsoome> meh taht readme is dumb. updated package list does show fc and emlxs packages are updated but change request descriptions dont mention those...[15:39:48] *** raichoo has quit IRC[15:40:20] *** raichoo has joined #openindiana[15:46:03] *** McBofh has quit IRC[15:48:14] *** ThomasB2k has quit IRC[15:48:48] *** mnaser has quit IRC[15:50:32] *** ThomasB2k has joined #openindiana[15:51:53] <sivanov_> ooooo just made loopback connection to the LUN via iscsi and ran iozone[15:52:06] <sivanov_> damn it loaded the server too heave[15:52:12] <sivanov_> heavy*[15:53:46] <sivanov_> was not bery good idea[15:55:04] <tsoome> its always good to start with basic tests;)[15:55:38] *** gervystar has joined #openindiana[15:56:10] *** McBofh has joined #openindiana[15:58:29] *** wonslung has joined #openindiana[15:59:33] *** ThomasB2k has quit IRC[15:59:56] *** ThomasB2k has joined #openindiana[16:00:23] <sivanov_> great, the test does not work:[16:00:58] <sivanov_> i created lun, connected to it via iscsi client to 127.0.0.1[16:01:16] <sivanov_> the target has tgp on 127.0.0.1[16:01:48] <sivanov_> found partitions in /dev/dsk[16:01:56] <sivanov_> run iozone on them[16:02:14] <sivanov_> AND the system was like weighed[16:03:28] <tsoome> what filesystem on client disk?[16:03:53] <sivanov_> FC4 config had no filesystem, raw disks[16:04:06] <sivanov_> FC2 - NTFS with 64k cluster[16:04:14] <tsoome> ah, you are running iozone on raw disk?[16:04:21] <sivanov_> yes[16:04:37] <tsoome> run on /dev/rdsk then[16:05:02] <tsoome> dsk is buffered io[16:06:41] <sivanov_> looks like i need: the system behaves weirdly: iozone runs write first, but i see tons of read in zfs iostat[16:07:13] *** tsoome1 has joined #openindiana[16:07:29] <tsoome1> what raid are you using on the pool?[16:08:51] <sivanov_> raidz : 3 vdevs of 5 disks[16:09:07] <sivanov_> now running on rdsk[16:09:09] <sivanov_> and[16:09:20] <sivanov_> get reads from pool[16:09:37] <tsoome1> well, to write on raidz, it needs to read whole stripe first[16:09:57] <tsoome1> then create parity and then write all stripe back.[16:10:21] *** tsoome has quit IRC[16:10:22] *** tsoome1 is now known as tsoome[16:10:38] <tsoome> she smaller the change, the more read (and write) penalty you get[16:10:38] <tsoome> s/she/the/[16:11:04] <sivanov_> data 9.14T 7.86T 978 878 122M 106M[16:11:04] <sivanov_> data 9.14T 7.86T 934 118 117M 537K[16:11:04] <sivanov_> data 9.14T 7.86T 590 2.13K 73.8M 263M[16:11:31] *** melliott has quit IRC[16:11:37] <sivanov_> but when i run from remote host i see only write activity[16:11:50] *** tsoome1 has joined #openindiana[16:12:08] <tsoome1> .oO bloody openvpn[16:12:33] *** melliott has joined #openindiana[16:13:06] <sivanov_> ok, i make 4 zvols @16GB each rather than one partitioned, and try without iscsi loopback[16:15:06] <tsoome1> if you have random writes with small blocks, you wanna keep your vdev as narrow as possible.... or use raid10 if there are too many writes (compared to reads)[16:15:13] *** skeeziks has joined #openindiana[16:15:33] *** tsoome has quit IRC[16:15:33] *** tsoome1 is now known as tsoome[16:19:59] <sivanov_> well, i get 20MB/s random read @64K block via iscsi[16:25:50] <sivanov_> found a nice asus server motherboard on ebay[16:26:22] <sivanov_> has 2x pci-e x16 2.0 and pci-e x 4 1.0 and pci-x[16:27:05] <sivanov_> probably going to upgrade home server[16:27:31] <sivanov_> to like one xeon 5472[16:28:33] *** Naresh has quit IRC[16:31:38] <kinesis_> supermicro x8dth-6f have 7 pci-e x8[16:32:49] <viridari> It's also USD$550[16:33:05] <viridari> yeah yeah... I know... server vs. desktop[16:35:14] <kinesis_> Jul 11 18:13:38 box idmapd[526]: [ID 388589 daemon.notice] Configuration unchanged[16:35:14] <kinesis_> Jul 11 18:23:38 box last message repeated 2 times[16:35:38] <taemun> xeon 54xx is two, verging on three generations old at this stage[16:35:44] <kinesis_> i have this notice every 10 minutes. how can i disable it?[16:37:54] <tsoome> check /etc/syslog.conf for *.notice ?[16:38:46] <kinesis_> tsoome, this notices only in /var/adm/messages[16:39:11] <tsoome> exactly.[16:39:44] <kinesis_> thanks, i see![16:39:55] <RoyK> erm - isn't gcc4 ins 151 yet?[16:39:58] <kinesis_> i think i can disable them only for idmapd[16:40:19] <RoyK> kinesis_: I have a few of those - works well :)[16:41:40] <kinesis_> RoyK, in my case i get smb_lookup_sid error ([16:42:13] <kinesis_> weekly now ([16:46:52] <kinesis_> i have this notices only on one server (with smb_lookup_sid errors), another one dont have such notices in log. syslog.conf is same! cant udterstand it.[16:47:05] <kinesis_> may me its key to my error?[16:47:25] *** raichoo has quit IRC[16:47:46] <tsoome> syslog is just for to show or hide messages;)[16:47:59] <tsoome> it has nothing to do with actual service[16:48:01] <kinesis_> yes, i know! )[16:48:41] <kinesis_> but 2 servers have same configs. one have this notices every 10 minutes. another dont.[16:54:13] *** lewq has joined #openindiana[16:56:50] *** kinesis_ has quit IRC[17:02:42] *** lkocman_ has joined #openindiana[17:11:00] *** Micr0mega has left #openindiana[17:19:33] *** edogawaconan has quit IRC[17:22:11] *** miine_ has joined #openindiana[17:24:27] *** miine has quit IRC[17:24:27] *** miine_ is now known as miine[17:25:02] <sivanov_> chages made pcitool, do the stay after reboot?[17:25:13] *** edogawaconan has joined #openindiana[17:25:28] *** Reepicheep has joined #openindiana[17:26:55] <sivanov_> taemun, this motherboard will sell probably for $150 and xeon 5472 is for 150/free shipping right now on ebay[17:27:27] <taemun> I'd rather have known unmolested hardware for a "server", but hey :P[17:27:44] <dkeav> picky picky[17:27:51] <sivanov_> yap ap[17:27:54] <sivanov_> *yap[17:30:01] <sivanov_> what can happen to a cpu in a server? where they cant overclock them[17:31:41] <sivanov_> again, about bying new hardware: my company bought ibm x3650 with 2.5" sas hdds, 3 servers, 18 drives, 7 drives died after two years of service[17:32:05] <sivanov_> AND in span of 2 weeks :)))[17:32:46] *** ivo_ has joined #openindiana[17:33:01] <dkeav> guess they don't mix their lots up huh[17:34:26] <sivanov_> AND they were in raids10, without damn ibm bloated serveraid agent in ESXes[17:37:29] <sivanov_> tsoome, btw iozone says there is 252MB/s read from 4 local zvols, so i lose about 100MBps between zvols-FC-client[17:42:21] *** dijenerate has quit IRC[17:47:20] *** nikolam has joined #openindiana[17:48:07] *** gervystar_ has joined #openindiana[17:48:27] *** gervystar_ has quit IRC[18:04:29] *** held has quit IRC[18:07:38] *** merzo has quit IRC[18:07:59] *** hajma has quit IRC[18:13:45] *** mnaser has joined #openindiana[18:17:05] *** mnaser has quit IRC[18:18:05] *** mikw has quit IRC[18:23:23] *** nikolam has quit IRC[18:23:46] *** ivo_ has quit IRC[18:24:41] *** DanaG has joined #openindiana[18:27:31] *** ThomasB2k has quit IRC[18:28:28] *** ThomasB2k has joined #openindiana[18:31:57] *** freedomrun has quit IRC[18:33:41] *** |AbsyntH| has quit IRC[18:35:17] *** lkocman_ has quit IRC[18:35:19] *** lkocman has quit IRC[18:35:34] *** xmikus01 has quit IRC[18:36:54] *** melliott has quit IRC[18:39:20] *** ivo_ has joined #openindiana[18:39:21] *** gervystar has quit IRC[18:39:33] *** melliott has joined #openindiana[18:41:30] <sivanov_> grrr 130MB/s read over FC4 and 400MB/s read from local /dev/zvol/dsk...[18:42:38] <tsoome> what you get from iscsi?[18:43:06] <sivanov_> iscsi_speed: http://www.pastie.org/2194966[18:43:38] <tsoome> with tuned tcp stack?[18:44:24] <sivanov_> only xmit/receive windows tuned to 1M[18:46:50] <sivanov_> can't try dual links & MPIO: this client PC has one etheret port died, other pc is taited with kaspersky[18:47:03] <sivanov_> and slower[18:48:13] *** melliott has quit IRC[18:49:11] <sivanov_> and iostat shows access to all 15 hdds[18:50:47] *** dijenerate has joined #openindiana[18:51:24] <sivanov_> im starting to think, that i get some bad case of slow RAM (this server can do about 1.2GB/sec only, according to memtest) and bad luck with cache misses/task switches across the CPUs[18:52:02] *** ivo_ has quit IRC[18:52:05] *** land0 has joined #openindiana[18:53:48] *** mnaser has joined #openindiana[18:53:48] <sivanov_> and opteron being NUMA can be really slow, passing data from devices to one cpu's ram then to another's cpu ram then back again[18:55:23] *** raichoo has joined #openindiana[18:59:37] *** syoyo_ has quit IRC[19:03:47] *** buffyg has joined #openindiana[19:03:47] *** ChanServ sets mode: +o buffyg[19:04:34] *** mnaser has quit IRC[19:04:45] *** syoyo_ has joined #openindiana[19:06:46] *** Botanic has quit IRC[19:07:29] *** syoyo_ has quit IRC[19:08:58] *** forquare has joined #openindiana[19:09:28] *** gea__ has joined #openindiana[19:11:52] *** Naresh has joined #openindiana[19:12:28] *** dijenerate has quit IRC[19:12:29] *** ThomasB2k has quit IRC[19:23:12] *** mnaser has joined #openindiana[19:23:23] *** mnaser has quit IRC[19:23:26] *** ThomasB2k has joined #openindiana[19:25:50] *** Botanic has joined #openindiana[19:25:50] *** Botanic has quit IRC[19:25:50] *** Botanic has joined #openindiana[19:27:19] <DanaG> Say, how do I make the smb support show dot files as hidden?[19:33:04] *** land_0 has joined #openindiana[19:33:10] *** land_0 has quit IRC[19:33:34] <sivanov_> ur hiding pr0n![19:34:26] <sivanov_> have not found that too[19:44:50] *** land_0 has joined #openindiana[19:44:57] *** land_0 has quit IRC[19:49:47] *** Alasdairrr is now known as AlasAway[19:51:50] <DanaG> No, I'm hiding .evince and all that crap.[20:00:20] *** DrLou has quit IRC[20:03:23] *** gber has quit IRC[20:05:54] *** freedomrun has joined #openindiana[20:06:58] <jkimball4> anyone built monotone scm by chance?[20:13:36] *** bens1 has quit IRC[20:13:46] *** skeeziks has quit IRC[20:19:41] *** land0 has quit IRC[20:21:23] *** TibshoOT has quit IRC[20:23:59] *** skeeziks has joined #openindiana[20:31:27] <DanaG> hmm, anyone know where to get Mono for OI?[20:32:08] *** joffe is now known as UnrealBorja[20:32:57] <DanaG> More specifically, I need mono-service2 and libmono-remoting[20:33:06] <DanaG> ah: http://www.mono-project.com/Mono:Solaris[20:33:34] *** UnrealBorja is now known as joffe[20:33:42] <DanaG> Okay, given these files, how do I install?[20:33:43] <DanaG> http://ftp.novell.com/pub/mono/third-party/codice/2.10/[20:33:56] <DanaG> there are .pkg.gz and -pkg files.[20:38:46] <richlowe> gunzip, pkgadd -d <the .pkg file>[20:41:08] <DanaG> Say, is Solaris supposed to boot 32-bit mode? Or how do you tell which mode it's booted in?[20:41:15] <DanaG> SunOS microserver 5.11 oi_148 i86pc i386 i86pc Solaris[20:41:33] <Woodstock> DanaG: isainfo -k[20:41:51] <DanaG> amd64. Cool. So uname is confusing me, then. =/[20:43:34] <Woodstock> an installed x86 solaris can always run in both modes, the bootloader grub usually decides which kernel to load[20:43:55] <DanaG> wow, I ran pkgadd on the extracted .pkg file, and now it's spewing garbage.[20:44:25] <DanaG> oh, wait, I did cat.[20:44:36] <DanaG> silly me and arrow-key command repeation.[20:51:34] <DanaG> weird, mono can't find libMonoPosixHelper.so[20:54:31] <RoyK> damn - I think I know the source to a rather nasty bug[20:55:12] <RoyK> A few times lately, I've seen a pool lose a drive, it gets replaced, I get an alert it's ok, and then, perhaps 15 minutes later, the box won't answer to even ssh[20:55:34] <RoyK> that's _after_ resilver is finished[20:55:41] <richlowe> what's the console say?[20:56:01] *** xmikus01 has joined #openindiana[20:56:22] <richlowe> also, putting 'set snooping=1' in /etc/system and hoping the deadman fires would be good[20:56:24] <RoyK> on the machine where it just happened, it's flooding the console with "disconnectind command timeout for target 23"[20:56:27] <richlowe> though it's been so long since I saw it work...[20:56:47] <richlowe> RoyK: mpt_sas?[20:56:51] <RoyK> "Disconnected timeout"[20:56:54] <RoyK> richlowe: yes[20:57:08] <RoyK> "Disconnected command timeout"[20:57:09] <RoyK> even[20:57:41] <RoyK> any idea what it might be?[20:57:59] <richlowe> Do you actually mean "mpt_sas" or do you mean "mpt, but SAS"[20:58:07] <richlowe> LSI are losers, so the driver names are hell :)[20:58:35] <RoyK> the driver is mpt_sas, lsi 1068 (3801/3080 boards)[20:58:44] <RoyK> 3801/3081 even[20:58:56] <richlowe> Oh, that's special[20:59:06] <RoyK> let me check - it might be just mpt[20:59:19] <RoyK> how do i check, btw? I've forgot[20:59:21] <DanaG> I see, it's looking in /opt/mono/lib/mono/2.0/ instead of /opt/mono/lib[20:59:34] <richlowe> RoyK: prtconf -D should show you, but it should also be part of the error messages you're quoting.[20:59:42] <RoyK> mpt[21:00:36] <RoyK> scsi: WARNING: /pci at 0,0/pci8086,340e@7/pci1000,30a0@0 dot .. (mpt0): Disconnected command timeout for Target 23[21:01:08] *** land0 has joined #openindiana[21:01:19] <richlowe> on the one hand, it's familiar, on the other it looks like a really common way for mpt to freak out[21:01:57] <RoyK> so would you say it's a driver or hardware failure?[21:02:10] <richlowe> I wouldn't place bets[21:03:12] <DanaG> oh, I typo'd LBIRARY[21:06:31] *** mikw has joined #openindiana[21:08:20] *** freedomrun has quit IRC[21:08:53] <RoyK> richlowe: https://www.illumos.org/issues/1197[21:09:07] <DanaG> Missing method get_Now in assembly /usr/LTSvc/LTSVC.exe, type Microsoft.VisualBasic.DateAndTime[21:10:06] <DanaG> oh, missing more libs.[21:10:44] <richlowe> RoyK: just saw, I don't know of any easy way to debug it. Presuming the complaints from the transport are real.[21:12:03] <RoyK> why should that happen just after resilver? on two machines?[21:12:13] <RoyK> quite different machines, in fact[21:12:27] <RoyK> well, controller's, the same, but that won't know much about resilver[21:15:14] *** radsy has quit IRC[21:15:16] <miine> RoyK: just curios: how did you solve it? Is the zp ool ok after rebooting?[21:15:42] <DanaG> so, Microsoft.VisualBasic.dll is missing.[21:15:46] <RoyK> miine: haven't rebooted yet - last I tried, it was, but started a resilver[21:16:39] <miine> RoyK: are you running something like smartmontools?[21:18:17] <miine> RoyK: but I don't think thats the drives fault (as SAS drives shouldn't have the Samsung F4 bug of hanging up when receiving a smartmon query while writing etc. ..)[21:18:19] *** ThomasB2k has quit IRC[21:18:25] <richlowe> RoyK: probably would be interesting, and worth adding to the bug, which disk t23 on that controller is[21:18:32] <richlowe> in relation to the disk failure and hot spare[21:18:36] <richlowe> the failed disk? the spare?[21:18:38] <richlowe> neither?[21:19:16] <RoyK> none of the ones that was resilvering, and neither of the spares[21:19:37] <richlowe> raidz? mirror?[21:19:39] <RoyK> c4t37d0 and c4t43d0 was replaced[21:19:41] <richlowe> and how full.[21:19:42] <RoyK> raidz2[21:19:50] <RoyK> 30%ish full[21:19:50] *** DontKnwMuch has joined #openindiana[21:20:02] <richlowe> there's a shitload of reports of mpt freaking out under heavy I/O in x4540[21:20:23] <RoyK> 11 7-drive RAIDz2 VDEVs[21:20:35] <RoyK> shit - this happens just then, under heavy load[21:21:05] <DontKnwMuch> my OI is idle so to say, smb chares and one nsf share, but no traffic, what I see in top is 85% idle and 15% kernel, what is using my cpu?[21:21:43] <RoyK> DontKnwMuch: no scrub running?[21:21:50] <miine> DontKnwMuch: intrstat ?[21:22:16] <DontKnwMuch> no, no scrub...[21:22:23] <RoyK> richlowe: what controllers would you recommend trying to try to get around this?[21:22:56] <richlowe> DontKnwMuch: dtrace -n 'profile-97hz { @[stack()] = count(); } tick-10s { printa(@); exit(0); }' is worth trying too[21:22:58] <richlowe> RoyK: I don't know[21:23:24] <DontKnwMuch> intrstat shows 12-15 for e1000g[21:23:31] <richlowe> DontKnwMuch: as a sarcastic answer to that kind of question, "top is using your cpu" isn't often _that_ wrong.[21:23:48] <DontKnwMuch> possible yes..[21:24:29] <RoyK> richlowe: top shouldn't spend system time, though[21:24:30] <DontKnwMuch> but not 15% of xeon 5530[21:24:49] *** melliott has joined #openindiana[21:24:58] <RoyK> DontKnwMuch: try mpstat - is one core very active?[21:25:17] *** land0 has quit IRC[21:25:38] <richlowe> RoyK: libproc is not particularly subtle, nor are reads of /dev/kmem reads.[21:25:40] <DontKnwMuch> yes, cor0[21:26:02] <DontKnwMuch> shows sys 40[21:27:22] <miine> DontKnwMuch: what NIC?[21:27:29] <richlowe> DontKnwMuch: lockstat -Is5 sleep 10[21:27:36] *** DrLou has joined #openindiana[21:27:36] *** ChanServ sets mode: +o DrLou[21:27:41] *** skeeziks has quit IRC[21:28:08] <miine> DontKnwMuch: sorry. jsut saw that you have an intel...[21:28:10] <DontKnwMuch> intel onboard, its 82576[21:28:45] <miine> so at least it's not the broadcom bge bug...[21:29:50] <RoyK> richlowe: how do I know which controller that was on?[21:30:21] <richlowe> I don't know how to easily map mpt0 to a slot. :|[21:30:33] <richlowe> perhaps cfgadm will hint at it, perhaps prtdiag will.[21:30:40] <RoyK> k[21:31:20] <RoyK> anyway - even if there was a hardware failure just after resilver, a single drive shouldn't really take down the box[21:31:22] <richlowe> DontKnwMuch: did you run the lockstat command?[21:31:31] <DontKnwMuch> yes, just now[21:32:22] <DontKnwMuch> It made my screen scroll off ... I did redirect it to a file, but now i have problems getting it to pastie or something like that :([21:32:36] <RoyK> try again :)[21:35:15] <DontKnwMuch> http://www.pastie.org/2198321[21:35:32] <DontKnwMuch> I dont know what to look for in this...[21:35:39] <miine> RoyK: I would say its the 3801 : http://opensolaris.org/jive/thread.jspa?threadID=114246[21:37:24] <RoyK> richlowe: apparently c4t23d0 died, and it's not alone - it's got to be a controller/driver issue - i have seen zfs kick off drives several times and found them live and well after testing on another system - this pool was flagged ok just five minutes before the system hung http://pastie.org/pastes/2198324[21:39:00] <RoyK> and the drives c4t37d0 + c4t43d0 had already finished resilvering[21:39:45] <DontKnwMuch> royk: uh., wow, a lot of drives[21:40:26] <DontKnwMuch> richlowe: I rebooted and it is still the same cpu useage..[21:42:10] <miine> RoyK: otherwise prtconf -D device should tell you...[21:44:03] <RoyK> miine: I have two controllers in this box, LSI 3801 and LSI 3081, same chipset, one with internal connectors, one with external ones[21:44:21] <RoyK> (don't remember which is which)[21:44:28] <RoyK> but c4 is on the external controller[21:45:06] <RoyK> I beleive I've seen the same issue with the internal one as well, though[21:45:13] <miine> RoyK: so it seems you have to open the machine or maybe the driver's log will tell you...[21:45:33] <RoyK> does the driver have a log?[21:46:58] <miine> don't know. but somewhere might be a message which card was detected... (I don't own a LSI controller so I can't say for sure)[21:47:03] <richlowe> DontKnwMuch: looks like power management[21:47:43] *** andygraybeal has quit IRC[21:47:58] <DontKnwMuch> bios setting? CPU powersaving thingly?[21:49:05] <DontKnwMuch> richlowe: I will go and disable it in bios, to see if it will help :) thank you for looking at it[21:49:06] <richlowe> I mean, that is where the system seems to be spending time[21:49:21] *** held has joined #openindiana[21:49:27] <richlowe> I'm not immediately sure why you're worried though.[21:50:37] <DontKnwMuch> I was just looking at the cpu load when copying 4T of files to compressed fs, and when I stopped, the CPU usage was as it was before, only a little change, therefore I thought what could this be, eating 15% of the cpu[21:51:44] <richlowe> looks really really idle in that output, beyond some small bits of IO[21:53:57] <DontKnwMuch> I was as a noob just looking at top, and kernel was 15%[21:53:58] <DanaG> weird, somehow I have a file named:[21:54:05] <DanaG> With a line break.[21:54:10] <DanaG> /^XÒ[21:54:11] <DanaG> ^H^B[21:54:28] *** kart_ has quit IRC[21:55:35] <miine> DanaG: maybe its more clear in UTF-8 ...[21:57:06] <DanaG> So anyway, is it just not possible to get the mono visual-basic stuff on OI, without major manual compilation?[21:57:25] <DanaG> I'm trying to use the Linux+Mono version of our management agent.[21:57:43] <DanaG> That's very low priority, though, considering the Linux agent itself doesn't do much.[21:58:27] *** sergefonville1 has joined #openindiana[21:58:34] <sergefonville1> good day[21:58:51] <sergefonville1> I'm trying to install uwsgi[21:59:05] <sergefonville1> and when I start it manually, everything is normal[21:59:06] *** jv_ has quit IRC[21:59:11] <sergefonville1> when I start it through smf[21:59:14] <sergefonville1> it errors[21:59:22] <DanaG> Oh, and this seems to be down. http://pkgsearch-repo.uk.openindiana.org/2011Q1/devel[22:00:54] <DanaG> oops, typo.[22:00:54] <DanaG> http://pkgsrc-repo.uk.openindiana.org/[22:01:26] <RoyK> richlowe: do you have links to other problems related to this board/chipset? you said there had been numerous reports...[22:01:39] <RoyK> or did you just google?[22:01:42] <richlowe> The links I know of would all have been b.o.o :([22:02:00] <RoyK> meaning...[22:02:06] *** jv_ has joined #openindiana[22:03:23] <alanc> they all point down the memory hole to the bit bucket now[22:04:36] <richlowe> google finds echoes[22:05:33] <richlowe> basically, variants on "Did a lot of I/O with <thing>, got <a lot of error messages like yours>"[22:05:53] <richlowe> bug states make them look uncared for, possibly meaning someone punted onto HBA or disk firmware[22:05:56] <richlowe> but I have no way to know.[22:06:35] <quasi> if the bugs are old, then perhaps http://wayback.archive.org/web/*/http://bugs.opensolaris.org[22:08:33] *** dijenerate has joined #openindiana[22:09:15] *** alanc has quit IRC[22:10:29] <RoyK> richlowe: I had similar issues with another box - swapped the controller with an LSI 92xx (IIRC 9201-8e or something) and the problems were gone - tried to switch back just in case, and after some time, the system started behaving like ADHD-like - I guess it's time to get some new controllers....[22:11:41] <richlowe> quasi: I've not had much luck with that :([22:13:09] <RoyK> interestingly, I tried to google some of the error codes and found hits for different platforms, linux and solaris alike[22:13:27] <RoyK> meaning it's quite likely the problem is in the hardware[22:16:02] <richlowe> In that instance, it's always best to blame firmware first.[22:16:31] <richlowe> my expectation is that hardware people understand hardware, and software people probably understand software. Firmware ends up with one set or the other trying to do the wrong job.[22:17:57] <RoyK> richlowe: I've tried several different firmwares - all the same[22:18:22] <RoyK> running latest IT firmware on everything now - no big difference[22:20:40] *** Guest98279 has quit IRC[22:25:21] *** gnomad has quit IRC[22:28:56] *** alanc has joined #openindiana[22:28:56] *** ChanServ sets mode: +o alanc[22:32:17] *** ThomasB2k has joined #openindiana[22:38:09] *** bens1 has joined #openindiana[22:40:19] *** kimc has joined #openindiana[22:50:05] <DanaG> say, why is ipp (illumos package something-or-other) using netbsd.org?[22:51:19] <richlowe> pkg-src[22:51:43] <DanaG> Anyway, the website shows perl5, but the local repo doesn't have it.[22:51:46] <DanaG> http://pkgsrc-repo.uk.openindiana.org/packages/perl5/[22:53:48] <tsoome> how many perls one can need...:P[22:53:54] *** gnomad has joined #openindiana[22:55:31] <DanaG> I need libwww.[22:56:26] <tomww> SFEperl-lwp.spec[22:56:41] <DanaG> Is that already packaged somewhere?[22:59:39] <tomww> eventually in herzen's testing repository[23:00:02] <tomww> but I'm not sure that is is already converted to hierarrchical package names.[23:00:11] <tomww> which OI build? 147/148 or 151 ?[23:00:21] <DanaG> 148.[23:00:42] <tomww> becasue the perl interpreter changed with 151, so on 147/148 the old SFEperl-lwp.spec is a "just works" with the old SVR4 short names[23:03:42] *** bens1 has quit IRC[23:06:18] <RoyK> DanaG: CPAN should work too[23:06:20] <RoyK> or may[23:13:21] *** forquare has quit IRC[23:18:47] *** ivo_ has joined #openindiana[23:23:45] <DanaG> What is CPAN? I can't seem to find what that is.[23:24:14] <DanaG> I google for openindiana CPAN, I get: "Problems using CPAN / Perl Modules with OpenSolaris (& OpenIndiana)"[23:24:20] <tomww> CPAN is pullution to the filesystem.[23:24:28] <tomww> *pollution[23:25:07] <quasi> search.cpan.org[23:25:32] <quasi> DanaG: or just run perl -MCPAN -e shell[23:25:50] <quasi> follow the instructions and start installing modules[23:28:12] <DanaG> thanks.[23:30:53] <DanaG> hmm, there's a whole boatload of "failed"[23:32:33] <herzen> DanaG: sorry, a dependency of SFEperl-lwp.spec doesn't build[23:35:14] *** pothos_ has joined #openindiana[23:35:24] *** mnaser has joined #openindiana[23:36:33] <sergefonville1> when I have spawn-fcgi create three forks, how do they share the same socket?[23:36:41] *** pothos has quit IRC[23:36:58] *** pothos_ is now known as pothos[23:38:38] *** raichoo has quit IRC[23:44:01] *** gervystar has joined #openindiana[23:44:05] <sergefonville1> it seems that when a process daemonizes it looses its contract[23:44:09] <sergefonville1> is that true?[23:47:14] <tsoome> afaik child does keep the contract.[23:48:54] *** sivanov_ has quit IRC[23:49:23] *** sivanov_ has joined #openindiana[23:51:42] <sergefonville1> I'm setting up uwsgi, and I see strang behaviour[23:51:51] <DanaG> That's one downside of OI: many fewer packages than, say, ubuntu.[23:52:27] *** hajma has joined #openindiana[23:53:42] <InTheWings> AlasAway: still relying on blastwave for libav anyway[23:54:14] *** nonot1 has joined #openindiana[23:58:20] *** dijenerate has quit IRC[23:58:47] *** gea__ has quit IRC