[01:04] stgraber: if i said to you that debhelper in trusty didn't seem to know how to do ifneq (,$(findstring $(DEB_HOST_ARCH_OS), linux)) while in xenial it did, would that ring any bells, or just be nonsense? [01:15] hallyn: doesn't ring a bell [01:16] kthx [01:18] i've let that block me too long, got other srus to do [03:34] hello, how do I make a SAS hard drive work on an Ubuntu desktop (not server) running on a mcahine without SAS (only SATA ports). What hardware do I need, and will the ubuntu desktop/kernel have builtin support or do I have to install additional software/kernel-modules? [03:41] qwebirc783: wouldn't you need a SAS controller card? [03:44] the other day I tried to find a usb<->sas bridge and couldn't find one for sale anywhere [03:45] nacc_: all i could find online was sas/raid server pages [03:45] an SAS HBA is your best bet; ubuntu supports a ton of them so whatever you buy will probably work [03:45] sarnold: that's most likely because no USB disks use SAS internally, so there are no chips for it [03:45] nacc_: ok, i'm not sure. if so, is that the only hardware I need - just basically use the PCIe slot on my desktop with a SAS controller PCIe card, and is it as simple as that... ubuntu desktop default kernel auto-detects card and drive as well? [03:46] JanC: yeah; there'd be near zero market for it.. I was curious if one existed for data recovery moments... [03:47] qwebirc783: if it is auto-detected on server it will also be on desktop [03:48] qwebirc783: you'll probably also need to buy some cables; it's not quite like sata you're used to.. this guide may help http://blog.zorinaq.com/?e=10 [03:49] qwebirc783: that would be my best guess, but i'm not sure -- SAS is pretty uncommon in the desktop, i think. As sarnold says you'll possible need the connecting hardware [03:49] qwebirc783: since most SAS controllers are used for lots of disks, they usually have 4, 8, or 16 local ports connected via 1, 2, or 4 actual connectors [03:49] qwebirc783: .. so you'll also need a cable that can go from the sff-8087 connectors to four individual disks [03:50] holy cow that blog post is almost six years old. [03:51] lol [03:51] JanC: right, but i haven't used ubuntu-server (even on my server machines) yet.. so, basically what you mean is if the card documentation says that it supports Ubuntu server, then it should also be supported under ubuntu desktop? is that because both server and desktop versions of ubuntu have the same builtin support for SAS? [03:52] they use the same kernel [03:52] i see [03:52] qwebirc783: yeah -- they use the same kernel. the server is almost identical -- except way fewer packages installed by default :) [03:53] there might be SAS controllers that need extra drivers, but guess most of these would come with some sort of linux drivers [03:54] yeah, i'm in the constant need for lots of inexpensive disk storage, and believe-it-or-not, the used SAS drives on ebay actually sell for a lower price (almost 33% less price on average) than SATA drives... so i'm thinking whether it is worth investing in the SAS controller card and if it would be as simple as the SATA i'm so used to... [03:55] qwebirc783: you might want to grab an SAS HBA with external ports, and pick up something like a JBOD enclosure.. [03:57] sarnold: most likely an USB-SAS bridge would require building a custom board with all the necessary chips (something like a SoC with USB & PCIE support + a SAS controller chip) on it + writing your own firmware (which could be linux...), at which point it's probably cheaper & easier to just plug a SAS controller in an old desktop/server or whatever :P [03:57] JanC: hah, yeah, I had muc hthe same train of thought :) [03:58] .. for the good and the bad. if it comes to it, _I_ could smack together something cheap off ebay to read my drives even if the Big Box's controller / expander died.. [03:59] qwebirc783: hey, sometimes you even find the things still have drives in them :) http://www.ebay.com/itm/48TB-Rackable-Systems-JBOD-w-16x-3TB-Hard-Drives-SAS-SATA-/360501845387?hash=item53ef959d8b:m:mqrzku8i20TtfaQTB9pQMcA [04:01] qwebirc783: I guess as SAS disks are mainly being used in servers, not any people want to buy them used & without a warranty [04:01] not many [04:02] and as they don't fit in regular PCs without extra hardware, that would make them pretty much "unsellable" :) [04:03] JanC: on the contrary theres loads of nutters who like to buy that kind of stuff used :) [04:03] maybe if they got some old server for free [04:04] got $150K USD? :) http://www.ebay.com/itm/52X-QUANTA-WINDMILL-OPEN-COMPUTE-NODES-4x-E5-2670-2-6GHZ-256GB-2x-250GB-W-RACK-/201426492935?hash=item2ee5f45e07 [04:05] LOL (just from the URL) [04:07] I've had my eye on this for a few months http://www.ebay.com/itm/IBM-LENOVO-THINKSERVER-RD650-2x-E5-2663-v3-10-CORE-192GB-12x-3TB-SAS-W-RAILS-/201468817323?hash=item2ee87a2fab -- but I think I've come down to the same decision you suggested -- I'm not sure I want to buy hard drives with an unknown amount of miles on them. [04:07] JanC: would http://www.amazon.com/SAS9200-8E-8PORT-Ext-Sata-Pcie/dp/B002QJZLCA/ be the only thing (along with the nessarry cables) I need for accessing SAS on my desktops? [04:08] qwebirc783: that may be a poor choice -- it only has external ports, which means it's designed to hook up to a JBOD sled [04:08] qwebirc783: if you guy one with internal ports then you can use a cable like http://www.amazon.com/Cable-Matters-Internal-Mini-SAS-Breakout/dp/B012BPLYJC/ref=pd_sim_147_3?ie=UTF8&dpID=51KIzZ%2BygyL&dpSrc=sims&preST=_AC_UL160_SR160%2C160_&refRID=1EXT3HTKQMMKVYTZENJR to hook up the drive [04:09] qwebirc783: of course if you -want- to buy the JBOD sled, then an external HBA is the way to go [04:11] sarnold: what's wrong with only having external ports - wouldn't that mean I could just use the appropriate SAS cable to connect it to individual SAS drives? [04:12] qwebirc783: I don't recall seeing any external-connector forward breakout cables [04:14] i don't mind if the SAS drives are outside the case if that's what you mean... I mean, even on my desktops, I have sata cables running outside of an unclosed case to SATA drives out in the open... so i'm not worried about the drives being outside the case, if that is what you mean [04:15] sure, I've done that too :) heh [04:15] but there are different connectors for internal vs external SAS HBAs [04:16] and I've seen plenty of internal forward breakout cables on different websites; I don't recall ever seeing any external forward breakout cables before [04:16] that doesn't mean they don't exist, it just seems unlikely [04:19] sarnold: can something like this work: http://www.newegg.com/Product/Product.aspx?Item=N82E16812710001 ?? [04:20] seems to good to be true, since the price is much lower... [04:21] qwebirc783: I -think- that kind of cable would let you attach SATA drives to an SAS controller. [04:22] jdstrand: aimed a q to you at end of bug 1393842 [04:22] bug 1393842 in libvirt (Ubuntu) "libvirt does not grant qemu-guest-agent channel perms" [High,Triaged] https://launchpad.net/bugs/1393842 [04:22] sarnold: not according to the customer reviews for that product: www.newegg.com/Product/ProductReview.aspx?PageSize=100&Item=N82E16812710001 [04:24] qwebirc783: I'm very skeptical. [04:28] hallyn: (a) eww (b) how did it ever work before? :) [04:29] just the sheer number of different connectors is intimidating: https://en.wikipedia.org/wiki/Serial_attached_SCSI#Connectors [04:30] qwebirc783: I completely agree [04:31] sarnold: yeah i dunno, i was looking at where it does an mknod and not finding it, so i'm not happy with it [04:55] sarnold: http://www.amazon.com/StarTech-18-Inch-Cable-Power-SAS729PW18/dp/B000V72AQ4/ The description for many of these cables says something like: "Allows you to connect a SAS hard drive to a SATA controller that supports SAS drives" . So, my question is how do I figure out which SATA controllers (which I presume really means motherboards and/or add-on cards) really support SAS drives? [05:05] qwebirc783: heh, I've never seen one that says it can; SAS can support SATA because there's a SATA Tunnelling Protocol of some sort that SAS HBAs and expanders are supposed to support [05:06] well, SATA & SAS both talk SCSI protocol to some degree [05:07] SATA uses a subset of what SAS supports, I suppose [05:07] yeah, it's entirely possible for a SATA controller to speak SAS. I've just never seen it advertised as a feature anywhere :) [05:07] time for me to bail, good luck qwebirc783 :) [05:08] thanks JanC :) [07:08] good morning [07:34] arges: so i'm confused. i've tried building xenial's libvirt in several places (and xenial and trusty), and 'debian/rules build' always fails for me at configure, bc ... it's trying to say --with-hal --without-udev [07:34] which means it thinks DEB_HOST_ARCH_OS is not linux [07:34] are you able to build it by hand? [07:37] hm, well, debuild -b gets further... am i going to have to change my ways? [08:12] coreycb, this was the problem for the puppet team: https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1550188 [08:12] Launchpad bug 1550188 in neutron (Ubuntu) "oslo_concurrency.lock_path: sensible default value is missing" [High,Triaged] [09:32] coreycb, I'm testing a generic approach to assuring that a default lock_path is set [09:33] thus avoiding patching config files... [11:38] is it intentional that xenial doesn't boot on i7 anymore? I get this "invalid opcode" trace http://paste.ubuntu.com/15204908/ when booting the current xenial cloudimage as Openstack instance on some part of my platform. wily is working fine in comparison. [11:48] ah, looks like this is a variant of https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1524069 , so I guess I need to find out how to work around this within nova [11:48] Launchpad bug 1524069 in linux (Ubuntu) "[Xenial] KVM trusty guest 3.13.0-68 raid6-pq panic in raid6_avx21_gen_syndrome() while probing grub devices [was: Xenial KVM: updating Trusty guest from 3.13.0-68 to 3.13.0-71 causes kernel exception]" [High,Confirmed] === wk-work is now known as whitekidney === tsimonq2alt is now known as tsimonq2 [12:41] jamespage, would it make sense for upstream to set a sane default? [12:41] coreycb, maybe [12:41] but its not really oslo_concurrency that needs todo that - each projects might be different [12:48] jamespage, lock_utils is generated based on code in oslo.concurrency though [12:49] coreycb, yes but it basically has no default [12:49] other than os.environ.get('OSLO_LOCK_PATH') [12:49] jamespage, but perhaps it should [12:49] coreycb, and what would it be? [12:49] we could do os.environ.get('OSLO_LOCK_PATH') or '$state_path/lock' [12:49] jamespage, not sure, maybe what it used to be [12:50] coreycb, exactly! [12:50] coreycb, this never ever generated a sane default [12:50] it was hard-coded by individual projects pre-generation, and regressed during move to generation [12:51] jamespage, would different projects have different defaults? [12:52] yes [12:57] hallyn: i'll take a look, smb last touched it, but when I reviewed it last i was able to build [12:59] jamespage, it looks like it's supposed to default to a tempfile if OSLO_LOCK_PATH isn't set: https://github.com/openstack/oslo.concurrency/blob/master/oslo_concurrency/lockutils.py#L49 === Algorithm is now known as ^King [13:03] arges, huh? [13:08] coreycb, ok so that's not in our current package I thinkl [13:08] coreycb, https://github.com/openstack/oslo.concurrency/commit/5021ef82fd8f0323b82d6d010bff9dab8a0cbcec [13:09] coreycb, only 22 hrs old [13:09] jamespage, ah, yes [13:09] coreycb, you wanna pick that for oslo.concurrency and I'll revert my neutron changes in git [13:10] jamespage, sure [13:11] smb: hmm [13:11] smb: i'm trying to reproudce hallyns issue with the currnet libvirt/xenial source [13:12] arges, since I (or rather bip) had some issues with certs on freenode I am missing the context about what problem [13:18] arges, trying something there again (may be gone here for a bit) [13:25] and he's gone ... [13:26] hallyn: i'm getting a failure in the test phase [13:27] hallyn: http://paste.ubuntu.com/15205543/ [13:36] coreycb, ddellav: designateclient break: [13:36] $ designate domain-list [13:36] ERROR: Exception raised: Could not find the requested resource: resources/schemas/v1/domain.json [13:38] coreycb, ddellav: https://bugs.launchpad.net/ubuntu/+source/python-designateclient/+bug/1550350 [13:38] Launchpad bug 1550350 in python-designateclient (Ubuntu) "missing resources" [Critical,Triaged] [13:40] coreycb, ddellav: git != upstream release tarball (again...) [13:43] jamespage, oslo.concurrency uploaded [13:43] coreycb, awesome thankyou [13:45] jamespage, what do you mean git != ? [13:45] coreycb, this is a debian pkg-openstack based package right? [13:45] yes [13:45] as in it does not use the upstream release tarball... [13:45] right [13:46] thus is missing information for python setup.py install [13:46] the files are there [13:46] ok. sigh.. [13:46] I'll take a look [13:52] jamespage, resources/schemas/v1/domain.json is in the orig tar ball but doesn't get installed for some reason [13:53] its because the source tree is not based on the result of a python setup.py sdist [13:54] jamespage, alright let me test with a uscan generated tarball [14:33] jamespage, you're right about designateclient. I can upload a new version with the upstream tarball. not sure how to get around this in the future unless we build everything with upstream tarballs instead. [14:33] in that case we'd have to not sync anything [14:34] coreycb, its one of the drawbacks of this way of working with git/debian [14:35] coreycb, re-uploading with the release tarball is ok imho - but you're right that it will just break on the next sync we doo... [14:35] jamespage, pita [14:35] coreycb, you can of course add some extra d/rules to deal with this set of missing file... [14:35] but that's a pita as well [14:36] jamespage, I'm still not clear as to why the file is missing [14:37] jamespage, and it must be broken in debian too, right? [14:37] coreycb, yes [14:37] jamespage, ok. I'll tell zigo about it. [14:41] jamespage, designateclient uploaded [14:44] coreycb, okies... [15:09] Hey guys [15:10] postfix keeps kicking me back the following error: fatal: cannot execute /usr/sbin/postconf! [15:10] and i have no idea what to do next, can somebody please point me in the right direction ;3 [15:11] hmm, properly install postfix [15:12] there is no right direction [15:12] cause we have no idea how you caused that issue [15:12] sure there is ... the opposite of the left direction [15:12] :P [15:12] patdk-wk: it only happens when i try to log into roundcube [15:12] are you *using* postfix? [15:12] any idea which logs i should check [15:13] yeah i am [15:13] well, I don't know anything about roundcube [15:13] but the fact it even cares about postconf at all, or postfix, is just seriously wrong [15:15] hmm i'm convinced its an issue with /etc/postfix/master.cf [15:15] Sky2939: take roundcube out of the picture. such an error can occur when postfix is restarted and you have a bad config file (like master.cf or whatever) [15:16] heh? then he should have clear log entries in /var/log/mail.log [15:16] so he should take a look [15:17] i've checked log entries [15:17] fatal: cannot execute /usr/sbin/postconf! is the error i receive in /log/var/mail.err and mail.log [15:17] hmm, that is the *only* log entry? related to postfix? in that log? [15:18] yep... unfortunately [15:19] and roundcube interface says "failed to connect to storage server" [15:19] well, that is cause your imap isn't setup [15:20] Sky2939: look over this carefully to try to get more detail into the problem: http://www.postfix.org/DEBUG_README.html [15:30] patdk-wk: if you manually run "/usr/sbin/postconf biff", what does that say? [15:31] "cannot execute /usr/sbin/postconf" sounds like something the shell would say... I don't think any part of postfix actually uses postconf, other than the packaging (init scripts, etc) [15:33] lamont, well, might be some roundcube script they ship [15:33] (ignoring the debconf path to configure the package altogether) [15:33] ogra_: true [15:34] lamont, initscripts use it [15:34] to setup the chroot [15:34] but I'd expect a complaint from postconf itself to take a different form [15:34] patdk-wk: exactly [15:34] I just cannot even guess where to begin [15:34] patdk-wk: if you manually run "/usr/sbin/postconf biff", what does that say? [15:34] it's just screwed at this point, and will likely take a lot of poking to find the cause [15:35] lamont, heh? [15:35] from a shell prompt, if you run that command, what is the output [15:35] biff = no [15:36] which tells me that something is probably trying to run postconf while in a chroot, which doesn't have a copy of the postocnf binary, or is missing shlibs that it needs [15:36] I dunno how you figured that out [15:36] I run my own postfix packages, and patches, I do not have a problem [15:36] well, "cannot execute" rarely comes from the binary you call ... but from a higher level [15:37] I'm going to leave you to these excellent people [15:37] (the shell ... or the linker) [15:37] lamont, try talking to sky2939 [15:37] oh. sigh. somehow, I thought you were the one with the issue... oops [15:38] Sky2939: if you run this from a shell prompt, what does it say: /usr/sbin/postfix biff [15:38] no, I am very close to postfix myself, and I don't really have a clue where to start [15:38] other than everywhere :) [15:38] cause it sounds like a deeper issue, that is likely affecting more than just postfix [15:38] patdk-wk: that's ebcause it's not a postfix issue, per se... it's above the code, in either some screwed up packaging, or such [15:38] yep [15:39] * lamont was just leaping to the postfix defense... [17:07] If I deploy a 16.04_server machine now using one of the daily/beta. When the distro becomes official if there anything else to consider other than upgrading from the repo? [17:07] if=is [17:10] hey folks [17:10] i have a custom interactive script i want to run automatically on login [17:10] via ssh [17:10] i always want scp / git over ssh to work [17:11] .bashrc can tell if its an interactive shell - can i throw the script at the end of .bashrc and let it run interactively [17:12] or is there a better way? [17:14] doesnt seem to work with .bashrc === magicalChicken_ is now known as magicalChicken === frediz_ is now known as frediz === mukhbiir is now known as neoark === Raboo_ is now known as Raboo === freeflying__ is now known as freeflying === seg_ is now known as seg === alexisb is now known as alexisb-afk === Azelphur_ is now known as Azelphur [17:23] hello, how do I make a SAS hard drive work on an Ubuntu desktop (not server) running on a mcahine without SAS (only SATA ports). What hardware do I need, and will the ubuntu desktop/kernel have builtin support or do I have to install additional software/kernel-modules? === Azelphur is now known as Azelphu === Azelphu is now known as Azelphur [17:24] qwebirc783, you need a sas port [17:24] sas and sata are NOT compatable with each other [17:25] sas can handle sata drives, if they support stun (I think almost everything does) [17:25] but sata cannot in any way handle sas at all [17:25] patdk-wk: from what I read only, i need a SAS controller [17:25] most popular would be an lsi hba [17:26] would something like this work: http://www.newegg.com/Product/Product.aspx?Item=N82E16812710001 ? === tyhicks` is now known as tyhicks [17:31] I actually have a server machine, but cannot confirm if it has a SAS port or not... it is supposed to have either a hardware RAID or software RAID which is supposedly meant to implement SAS [17:31] I cannot see any SAS cables sticking out of the power supply. [17:35] will any of the SAS controller cards (most of which seem to be PCIe cards) work to provide me access to SAS hard drives? === _fortis_ is now known as _fortis [17:36] arges: fascinating [17:37] arges: so can you build by doing "debian/rules build && fakeroot debian/rules binary"? [17:37] your test failureslook like they might be due to building in sbuild or something and not having a pty? dunno. [17:37] The test failure i was originally looking at was jamespage's error in trusty, which I now also get, in tests/xlconfigtest.c [17:38] smoser: ^ [17:38] my server's model number is: 70B5001TUX . Can someone please verify if it supports SAS without any additional hardware (or if additional hardware is needed , what would be necessary)? [17:38] but fwiw even in xenial 'debian/rules build' fails. I'm wondering why. I've only ever had that fail for me on kernel [17:39] qwebirc783, it doesn't support sas [17:39] you need to install an sas pcie card into it [17:39] hallyn, i think i'm missing some backscroll [17:40] patdk-wk: how did you find that out? according to this page, it does: http://www.cendirect.com/main_en/tech_specs_JO48730X.html?rPart_no=JO48730X [17:40] why are you quoting some random page? === tarpman_ is now known as tarpman [17:43] oh it might [17:43] stupid lenovo naming [17:43] at the bottom of the motherboard, does your model have two sff-8087 ports? [17:43] patdk-wk: well, i'm quoting some page simply for the sake of quoting some page.. i just do not want to spend additional money on additional hardware if I did not need it in the first place. Also, it would seem like false advertising for a business/vendor to state something false about a product that it sellss.. [17:43] http://www.tweaktown.com/image.php?image=imagescdn.tweaktown.com/content/6/6/6669_17_lenovo_thinkserver_td340_tower_server_review_full.jpg [17:43] *not quoting* [17:43] see in that picture, 6 sata ports (one is blue) [17:44] and then next to it with those big black wires going into it, is the sff8087 ports (sas) [17:44] those use the onboard intel scu crap though [17:46] smoser: wtf. i swear i typed 'smb' [17:46] smoser: ignore [17:46] patdk-wk: wow, i see that cable on that motherboard now... so does that mean it does support SAS? sorry, i'm a total noob to SAS, so if so please explain what I would need to make it work? [17:46] those are sas ports [17:47] they are via the intel scu controller though [17:47] and support for that should exist in linux/ubuntu [17:47] but I have never used it [17:47] hallyn, command line complefusion at your service [17:47] so, give it a try and see, you might need cables, dunno [17:47] those cables on my motherboard are tucked tightly underneath the front bays where I cannot even access the front part, so I cannot even see wht the front end of that cable looks like to verify or not if that is SAS... [17:48] it's sas [17:48] 4 drives per cable [17:48] smb: how do yo ubuild libvirt locally? [17:48] plug the drive into a slot connected to those [17:48] hallyn, ubuild? I just use sbuild on the package files [17:49] hallyn, dpkg-buildpackage -D [for deps] [17:49] hallyn, dpkg-buildpakage -b -uc -us [17:49] hallyn, that should work in the unpacked tree [17:50] smb: it bugs when debian/rules build doesn't work, [17:50] hallyn, yeah though that is basically used by all methods above and it works for me [17:51] smb: can you retry with a fresh libvirt tree? [17:51] bc it fails for me, everywhere [17:51] it mis-detects the host os and uses build flags for no-linux [17:51] then dies bc it can't find hal [17:51] hallyn, fresh from where? [17:51] xenial [17:51] pull-lp-source libvirt [17:53] hallyn, installing dependencies... [17:54] patdk-wk: I cannot plug any SAS drives yet. i was considering purchasing some used SAS drives on ebay, since they are, believe-it-or-not, lower in price than the SATA drives! i'm constantly in the need for inexpensive storage, and wanted to look for a cheaper alternative... just to pull up an example off of ebay, will these hard-drives work with my 70B5001TUX motherboard?: [17:54] http://www.ebay.com/itm/4X-Seagate-Constellation-ES-2-ST33000650SS-3TB-7-2K-64MB-SAS-6Gb-s-3-5-HDD-/131732598115?& [18:00] hallyn, ok, build started. Though one difference maybe, I am using a xenial chroot on trusty [18:01] smb: it fails for me even just on a native trusty vm or container, and on my xenial laptop, so i expect it'll fail just fine [18:01] hi guys my laptop overheated shut it self down, now when i want to boot it up again it hangs on the init ramdisk [18:07] hallyn, first round of tests passed... [18:07] smb: no way! [18:08] smb: lemme start a digitalocean droplet we can share === Sling_ is now known as Sling [18:09] hallyn, wonder whether we could do something like this on a porter [18:11] patdk-wk: I now see that that cable is connected to some kind of SAS controller hardare that has what-seems-to-be 4 SAS ports sticking out from the front bay. would this seem correct for use with SAS drives? [18:11] smb: ssh ubuntu@04.131.129.237 with your lp ssh key , do tmux -a ? [18:13] hallyn, is that really 04. ^ [18:13] ok no [18:14] d'oh [18:14] 104 [18:18] hallyn, so fwiw the local build finished ok === Piper-Off is now known as Monthrect [18:21] Hi I am curious about upgrading a raid5 config that uses 4 2tb drives on a Dell PERC6i card. I was wanting to add 4 more 2TB drives to the raid or having some way of associating two raids together somehow. Iirc to actually add drives to a raid you need to back everything up first then delete the raid and make a new one with all the new drives? Using Ubuntu 14.04.3 LTS and was just looking for some suggestions! :D [18:40] it looks like I have 4 SAS ports on my server machine. can someone confirm to me if SAS drives can be used like regular SATA hard drives for storing data without RAID? [18:41] and do I have to plug in 4 SAS drives for the SAS to be accessible, or will one be sufficient? and if I do plug in 4 at a time, will GNU/Linux OS see it as 4 individual 3TB drives or 1 big 12TB drive? [18:44] qwebirc783: hey :) sounds like you're still having fun :) [18:45] qwebirc783: it depends if you have a raid controller or "it mode" controller -- the raid controllers are usually more expensive [18:45] sarnold: ;-) found out that my server machine actually seems to have SAS ports :-o [18:45] qwebirc783: that's great, saves a pile of money :) [18:47] yep, i'm hoping to spend ~$14 per TB on used SAS drives versus $18 per TB on used SATA drives on ebay.. I spend about $5000 in SATA HDD last year, so that would be significant savings... ;-) [18:48] oh my :) [18:52] there are screenshots of my motherboard's BIOS setup posted here with the SAS configuration showing... can you confirm if i'll be able use the 4 SAS drives as 4 individual data drives showing up under GNU/Linux as devices /dev/sd{a,b,c,d} ? [18:52] sarnold: http://www.tweaktown.com/reviews/6669/lenovo-thinkserver-td340-tower-server-review/index3.html [18:53] qwebirc783: you may need to change the "storage oprom policy" -- a coworker reported some issues with thinkservers controllers but he was able to get it to work [18:54] qwebirc783: the drives may have different letters, but that ought to work well [18:56] sarnold: right, I prefer to manage each drive manually as individual devices without software/hardware RAID and/or LVM. so, each drive having its own [18:57] its own FS (like ext4 or some other...) [19:00] qwebirc783: yeah, I'd rather have an IT mode firmware too -- but because I'd rather run zfs and let them handle the hard work :) [19:07] sarnold: i've never used zfs before but, will '/sbin/mke2fs' work the same way with SAS drives as it does with SATA drives? and will the SAS drives show up as /dev/sd* or /dev/scsi* or ??? [19:08] qwebirc783: yeah, the filesystems just need a block device; probably you won't even notice a difference === alexisb-afk is now known as alexisb [19:12] you have no raid on that sas controller [19:12] or rather, it's raid assisted booting stuff [19:12] all raid for those disks would be *software raid* [19:13] used to be dmraid in linux would do it, but mdadm took it over [19:13] without touching it, they should all appear as seperate drives [19:20] that's good to know. SAS all the way from now on.. (or until i'm convinced otherwise) [19:27] some more questions about my 70B5001TUX server: the specs say that it supports (1.35V / low-voltage) DDR3L RAM up to 1600. Does that mean it will support (1.5V / regular-voltage) DDR3 <=1600 RAM? [19:28] probably best to check the manual on that [19:29] depends [19:29] normally yes, but lately no [19:30] atleast all the laptops I have, can only use ddr3 (1.5v) or ddr3l (1.3v) and you cannot mix them up [19:30] servers, normally you can, but check first :) [19:32] manual does not state anything about wheter or not regular-voltage DDR3 is supported. it only states that low-voltage DDR3L is supported. [19:35] must be ddr3l [19:35] normal 1.5v will not work [19:35] maybe [19:35] what is in it now? ddr3l? === masACC is now known as maswan [19:37] yes ddr3l now. but since ddr3 is cheaper, was thinking about adding ddr3 rather than ddr3l [19:37] it says explicitly you cannot mix [19:37] they all must be the same [19:38] i thought you said: "servers, normally you can" [mix] ... ?? [19:38] I didn't say mix [19:38] normally install one or the other [19:39] oh i see.. now on ebay some 184-pin ddr3L are listed but they say they are intended for Dell servers.. would they work only on Dell servers and not on my Lenovo ThinkServer? [19:39] * patdk-wk also loves the fact, when the manual shows, installing a new ethernet card [19:39] then has a picture of a lsi 9211-8i being installed [19:39] patdk-wk: haha [19:41] qwebirc783, it doesn't matter if it says for dell or lenovo or whatever [19:41] as long as the voltage, and rank and speed are the same, your good [19:41] you likely have all rank2 or rank1 [19:42] I would just look for ddr3l r2 and same speed [19:42] oh ya, do make sure if your using ecc registered ram, or unbuffered [19:42] if it's udimm's, it's just annoying personally :) [19:43] I want to use only RDIMM since that way I can capitalize on the max-machine-allowed memory. [19:44] so does the same thing you said apply to RDIMM as well as UDIMM? [19:44] everything applies to both [19:45] just udimm is ultra annoying [19:45] and expensive if you need udimm ecc, and limited ram ability [19:45] i see. then, does it matter if one of the memory sticks already installed in the system is 8GB and the rest that I will be buying are 16GB, or do all of them have to be the same size as well? [19:45] but it is *faster* [19:45] well, according to the manual, same size [19:45] but it should work fine [19:45] wild shot in the dark but has anyone ever installed IBM's ITM agent on Ubuntu? [19:45] some CPUs/mobos support mixes, some don't [19:48] patdk-wk: so wait, because the memory installed is 8GB (DDR3L) module, then the 16GB (DDR3L RDIMM) modules I purchase will not work with the 8GB? [19:48] qwebirc783, the manual says no [19:48] I say, most likely it will work [19:48] where did you d/l manual from? [19:49] http://content.etilize.com/User-Manual/1027666706.pdf [19:49] page 97 [19:50] now what you have to do, if you do mix them [19:50] is only one type per bank [19:50] you have 6 banks, 3 per cpu, each using a pair of memory modules [19:51] it *should* be fine to install 2 8gigs, and 2 16gigs, in banks A and B and work perfect [19:51] unless lenovo did something really odd [19:51] but I can tell you, it works fine for all my HP, DELL, supermicro stuff [19:52] and since memory controller is built into the cpu, it should not be an issue at all [19:52] as long as you stick to what the cpu needs, instead of what the motherboard says [19:53] those manuals do take the most strict case, of all cpu's they support [19:53] instead of what your specific cpu can do [19:54] i see. CPU #1 installed is: Xeon E5-2403 v2 . what you said about different sizes should most probably apply to it? [19:55] (default CPU installed by vendor, btw) [19:55] yes, it should be fine with different sizes [19:55] in fact default CPU for that model installed by lenovo itself [19:56] though, I can't guarentee you 100% as I don't have any lenovo servers/desktops [19:56] but I will be 99% confident it would work [19:56] though, I do also have plunty of 4g/8g/16g sticks to test with first [19:58] qwebirc783: http://ark.intel.com/products/75975/Intel-Xeon-Processor-E5-2403-v2-10M-Cache-1_80-GHz [19:58] qwebirc783: looks like intel says that processor tops out at ddr3 1333 -- and it doesn't have the Intel Flex Memory Access support === rmc3_ is now known as rmc3 [19:59] oh, that flex memory is new [19:59] different sizes WITHIN the same bank [19:59] OH [19:59] I normally put that down as, not a good idea at all [19:59] yeah it seems crazy [20:00] no, what was crazy [20:00] so I assumed it meant different sizes / layouts in different banks :) [20:00] was when I wasn't paying attention [20:00] and install two different cpu's, I think a E5640 and a E5620 [20:00] on the same motherboard [20:00] and it booted [20:00] o_O [20:00] then I found out intel supported that now [20:00] !!! [20:00] holy cow [20:00] it declocks the faster cpu to the lower one [20:01] but you still get the extra cores atleast [20:02] hmm, so using same capactiy per bank, it runs symetric, full speed [20:02] if you mixmatch with a bank, it goes asymmetric and slows down [20:02] but still faster than 1 stick [20:02] that xeon CPU doc says DDR3 is supported... but i guess motherboard will not let you mix DDR3 with DDR3L. but it says flex memory is "no"... first of all, what is "flex memory" and what does it mean that my Xeon does not support it..? [20:02] don't worry about flex [20:02] you have 3 banks per cpu [20:02] in each bank (2 memory slots) install same size [20:03] flex lets you install different sizes i nthe same bank, that used to be a no-no [20:03] so basically just assume [20:03] always order memory in pairs :) [20:03] and your good [20:03] I still wouldn't mix and match ddr3 and ddr3l but you probably can [20:04] and why does my xeon spec not list DDR3-1600... it only goes up to DDR3-1333, does tht mean if I buy DDR3L-1600 for use with this CPU, it will only run at DDR-1333 speed? [20:04] the thinkserver spec clearly states up to DDR3L-1600 [20:05] in other words, i do not want to spend extra $ on DDR3L-1600, if CPU is only capable of speeds up to DDR3L-1333 [20:05]