[00:50] What alternatives to MAAS + juju are there with Ubuntu? For installing and maintaining a cluster of Ubuntu nodes. [00:53] mgi: A traditional preseeded PXE install, handing over to something like Puppet? [00:53] (Which by no means is an Ubuntu specific answer.) [00:53] Yeah, that's fine, I'm not after Ubuntu specific, just Ubuntu-enabled :) [00:53] (as in, I'd rather not use rocks) [00:55] So, does Puppet do the actual install for you? [00:55] (just reading the site now) [00:55] mgi: No, Puppet needs an existing (basic) os install to run off. [00:56] So I need to deploy Ubuntu (or whatever OS) to all the compute nodes first? [00:57] mgi: Yepp, for example using a preseeded install [00:57] https://help.ubuntu.com/12.04/installation-guide/amd64/appendix-preseed.html [00:57] Ah brilliant, I think that's what I was looking for. [00:58] I've never done preseeding before - and will read the docs now - but before I get too excited, is this a valid approach for installing up to 80 nodes? [00:59] mgi: Can't see why not. Even with a limited number of server it is still nice to have the install automated, making it easy to reinstall broken nodes, etc. [00:59] Yeah, cool. I guess my other option is to install them all using MAAS and then use Puppet after that. [01:00] mgi: Well, even with a preseeded install I think you will just want to use the preseeds for the basic, and then hand of to Puppet as a final stage. [01:00] Ok, sounds good. Thanks! [01:01] mgi: Both because it is easier to do a lot of the detailed configuration in Puppet, and because it is a good thing to have as much as possible controlled by Puppet, when you need to roll out configuration changes on existing nodes. [01:02] Yeah, Puppet sounds like a good way to manage anything beyond the bare-bones install. [01:11] New bug: #1023729 in maas "maas-import-isos requires wget, but maas has no wget dependency" [Undecided,Confirmed] https://launchpad.net/bugs/1023729 === cpg|away is now known as cpg === bobble_ is now known as bobbow === Guest42446 is now known as exocaesar === n0ts_off is now known as n0ts === cpg is now known as cpg|away === cpg|away is now known as cpg === n0ts is now known as n0ts_off === cpg is now known as cpg|away === dendrobates is now known as dendro-afk [05:16] hallyn: it's late here so I won't investigate until tomorrow, but it looks like containers spawned through the API fail to reboot (they just die) === cpg|away is now known as cpg [07:08] does ubuntu server comes with LAMP plus ssl? and send and recieve mail functions? === cpg is now known as cpg|away [07:27] hi all. I have a home server that used to have a mirrored root, but only has a single drive now (one of the two). This means the mirror is "broken" - how can I remove the missing drive from the mirror to shut up mdadm monitoring? http://paste.ubuntu.com/1094469/ <-- this is how it looks today [08:05] RoyK: it's rebuilding right now [08:05] RoyK: look at mdadm --fail [08:05] example: mdadm --fail /dev/md0 /dev/sda1 [08:06] ivoks: that works well if you have a device name to give mdadm - I don't [08:07] then replace the disk, it's already marked as failed [08:07] I don't want to replace it [08:07] oh, now i get it [08:07] i want to kill raid [08:07] no, not kill the raid [08:08] just reduce the number of mirror copies to one [08:08] linux supports "mirrors" with only one side [08:08] mdadm /dev/mdX --grow --force -n 1 [08:10] whee - thanks! [08:10] then all I need to do is find the missing spare in md0 [08:16] Jamespage, ping [08:16] babyface_, morning [08:17] jamespage, good morning, James [08:17] jamespage, how about the tomcat bug? [08:17] Jamespage, when will it be fixed? [08:18] babyface_, so I noticed - although this is actually a test failure rather than related to last weeks issue with MIR's etc... [08:18] I will look at it now [08:18] Jamespage, ok, thanks. === cpg|away is now known as cpg [09:03] babyface_, hey - so I could not reproduce the issue seen in the daily ISO test - I re-ran and it cleared itself [09:03] babyface_, this may actually point to a bug; but I'm not sure what it is yet [09:03] it looks like the control port on 8005 for tomcat7 is not listening when the test executes.... === cpg is now known as cpg|away [09:20] babyface_, I notice that the amd64 version was also doing it three days OK - but has also cleared... [09:20] odd === almaisan-away is now known as al-maisan [09:26] stgraber: ping [09:30] apw, smb: where can I discover which modules are built for arm/omap4? === n0ts_off is now known as n0ts === railsraider_ is now known as railsraider === jtv1 is now known as jtv [10:34] jamespage: there is another channel, #ubuntu-arm, that might be appropriate for arm questions [10:35] RoyK, I'll try there instead... === n0ts is now known as n0ts_off [10:57] anyone here tried to setup kvm on multiple hosts with shared storage? [11:06] * RoyK gives up and installs CentOS :þ [11:12] hrm [11:13] anyone with an idea of why the iscsi initiator on these servers fail to connect to their LUN on bootup, but why it works just after reboot with a simple iscsiadm -m node -l ? === matsubara is now known as matsubara-afk [11:22] on which servers? [11:23] RoyK: i guess you need to tell the initiator to log in upon bootup? [11:26] pmatulis: normally it should [11:27] pmatulis: I've setup this before, similar setup, but other targets, and once setup and logged in, it should do so automatically on bootup, but something may be timing out - dunno [11:27] we're not using any fancy authentication, just matching an autogenerated target name [11:27] is mounting done before iscsid is started? [11:28] it really shouldn't [11:28] but I don't know upstart that well [11:28] it's just basic 12.04 === n0ts_off is now known as n0ts [11:30] so, it's not /-on-iscsi? [11:30] no, root is on a local mirror [11:31] on both nodes === n0ts is now known as n0ts_off [11:31] but they're supposed to share a volume on a SAN with OCFS2 for VM storage [11:31] and, once system boots, you can easily mount the the target? === railsraider_ is now known as railsraider [11:31] after I've logged into the lun with the above command, but it doesn't show up before that - but w8... think I found something [11:33] Which MTA do you recommend for a server that only sends email? so that cron, for example, can send emails? [11:34] !postfix [11:34] postfix is the default !MTA and !MDA on Ubuntu. For help, read https://help.ubuntu.com/community/Postfix and https://help.ubuntu.com/community/PostfixBasicSetupHowto - See also !MailServer [11:34] Thanks. [11:36] ivoks: hrmf - seems node.startup in /etc/iscsi/iscsid.conf was set to manual, by default. However, changing it to 'automatic' didn't change anything - the iscsi LUN isn't in /proc/partitions after bootup, and really, it should be [11:39] newbie. Can someone explain how to change Ubuntu Server to a static ip address, the default installation set it to dhcp. [11:40] nokia: edit /etc/network/interfaces - set it to static - google that for more info ;) [11:40] !interfaces [11:41] RoyK every time i reboot my asdl router all MY PC's Laptop's and Server ip's change. How do i make them all have same STATIC ip address's. [11:52] RoyK: i remember having to configure each lun individually for automatic login [11:53] hrmf - seems it now works on node1, but not on node2, configured the same [11:54] iscsiadm -m discovery shows my sendtarget, iscsiadm -m node shows my target '172.31.1.67:3260,1 iqn.2001-05.com.equallogic:0-8a0906-79b9aec05-128001a0ef84ffff-kvm-test-vol1', but it's not logged in [11:55] hm... and dmesg tells me eth1 was started ~3 seconds after iscsi [11:56] * pmatulis facepalm [11:56] ? [11:57] pmatulis: I didn't set the order... [11:57] well, network is needed before iscsi right? [11:57] I would think so... [11:57] so... smells like a good old byg [11:58] so it was a 'buntu facepalm, not a RoyK facepalm [11:58] :) [11:58] and i've seen this kind of bug before, so prolly a regression. but no one else running iscsi on 12.04? [11:59] node.conn[0].timeo.login_timeout was set to 15 - trying 60... [11:59] ah ok [12:02] these dell machines spend like 3-5 minutes just to POST [12:04] duh [12:04] increasing timeout to 60 didn't help [12:05] good morning [12:06] New bug: #1006738 in php5 (main) "php5-fpm segfaults with error 4 in libc-2.15.so" [Low,New] https://launchpad.net/bugs/1006738 [12:07] pmatulis: i have iscsi on 12.04 [12:08] but i set networking within initramfs, so i doubt that counts :) [12:08] bingo [12:08] :) [12:08] but [12:08] iscsi is restarted once network interface is up, so that shouldn't be a problem [12:09] if you look at /etc/network/if-up.d/, you'll see iscsi bits and pieces [12:09] it's a link to /etc/init.d/open-iscsi, iirc [12:10] questions are... do you have multiple interfaces? does restarting open-iscsi works once system is booted? === al-maisan is now known as almaisan-away [12:17] hrmf - bug 1025240 [12:17] Launchpad bug 1025240 in open-iscsi "iscsid is started before networking" [Undecided,New] https://launchpad.net/bugs/1025240 [12:17] RoyK: your bug is invalid [12:17] RoyK: open-iscsi is restarted after the interface is set up [12:18] the fact that iscsiadm works is a proof [12:18] iscsiadm wouldn't work if iscsid isn't running [12:19] hm... then why doesn't it connect to its LUN? [12:20] why do you think it's not connected? you don't see /dev/sdX? [12:20] nothing in /proc/partitions [12:20] which is a rather good indication ;) [12:20] if I login manually, it shows up there [12:21] New bug: #1025239 in ipxe (main) "kvm-ipxe does not support https" [Undecided,New] https://launchpad.net/bugs/1025239 [12:27] so, concentrate debugging on 'why isn't discovery done automatically' [12:27] not on 'iscsi is started before networking' [12:28] I am [12:28] but discovery looks ok [12:28] everything is reported correctly [12:28] but it doesn't login to the target [12:33] ivoks: heh - check the bug again :-þ [12:33] there you go [12:35] I just wish the installer would stop changing the iscsi name so many times during the install [12:36] ivoks: it would be rather nice if that stuff was documented, though === Pupeno_W_ is now known as Pupeno_W [12:46] hey folks I setuped ubuntu server and openerp in it, I watn somehow to access it from remote distance, not sure how to do it, for example as I have wifi on same internet, and I go to the ip and port it simply doesn't show anything [12:46] any udw [12:46] any uide [12:46] ideas"" [12:48] stgraber: the reboot logic was wrapped around the core start, so that's conceivable. [13:11] New bug: #724616 in ceph (main) "not lintian clean" [Wishlist,Triaged] https://launchpad.net/bugs/724616 === zyga is now known as zyga-food [13:21] hi, I'm not sure this is the channel, but I will ask you: I use Ubuntu 12.04 server for host and guest and have QCOW2 images for the guest. The problem is that with initial "virtual" size of 10GB, now the image is with physical/disk size of 15GB. This is causing me out of space problems all the time. Can anyone tell me if this behaviour is normal. How should I limit the file size of the image??? [13:22] You can have a look here for more info: https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1025244 [13:22] Launchpad bug 1025244 in libvirt "qcow2 image increasing disk size above the virtual limit" [Undecided,New] [13:26] New bug: #961389 in txlongpoll (main) "[MIR] txlongpoll" [High,In progress] https://launchpad.net/bugs/961389 [13:30] hallyn: yeah, I guess that makes sense. I'm kind of surprised I haven't tried it before though :) How much pain will that be to fix it? [13:30] hallyn: I don't have anything relying on reboot in those containers yet, but that probably won't be the case for long [13:39] stgraber: not sure. lemme finish verifying libvirt on precise sru then i'll take a look [13:39] sure, no hurry. thanks === zyga-food is now known as zyga-food] === zyga-food] is now known as zyga-food === dendro-afk is now known as dendrobates [13:58] hm, just updated from precise-updates to precise-proposed, and ssh now takes a loooot longer to get in === zyga-food is now known as zyga [14:00] hallyn: check dns resolution still works [14:00] hallyn: the most common reason for slow logins is broken dns resolution of the client from the server [14:00] reverse dns [14:00] Daviey: it doesn't seem to, but i don't understand why [14:00] doesn't seem to be part of what was updated [14:02] oh. heh. maybe i see the problem. [14:10] jamespage: hey did you check openvswitch on arm? === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === cedr` is now known as cedr [14:38] zul, won't work yet as openvswitch module is not built with the kernel [14:39] thats what i thought [14:39] zul, hit the same thing with ceph rbd today as well BTW [14:39] rbd is also not built as far as I can tell [14:49] zul, actually that was only the case for omap4 - its enabled for all other ARM targets [14:52] so highbank should work? [14:56] New bug: #1020603 in quantum (universe) "[MIR] quantum" [High,New] https://launchpad.net/bugs/1020603 === mrevell_ is now known as mrevell [15:12] anyone here? [15:14] if a bug gets marked 'invalid', it doesn't show up in LP searches right? [15:16] hallyn: correct, but you should be able to search in an advanced search for it [15:18] hallyn: what did you have to do for libvirt permissions? [15:19] zul: I had to chown /var/lib/libvirt/images to libvirt-qemu:kvm and make it group readable [15:19] but i notice that on precise it isn't like that either [15:23] Hi guys. We had ldap auth via ssh on our server, but now want to move back to "normal" ssh authentication. Any odeas whcih files I need to change? [15:24] adac: etc/pam.d/sshd ? [15:25] zul: right now it seems to be working for me. [15:25] th0mz, I thougt maybe: /etc/pam.d/common-auth [15:25] but I'm unsure [15:26] th0mz, curretnly looks like this: [15:26] cat /etc/pam.d/common-auth [15:26] auth sufficient pam_ldap.so [15:26] auth required pam_unix.so use_first_pass nullok_secure [15:27] zul: ah i think it just needs go+x [15:29] zul: I'm guessing libvirt-bin.postinst no longer does chmod 711 ${ROOT_DIRS} (checking) [15:31] hm, it says it does [15:34] Problem in setting up openldap? Getting this errro :ldap_bind: Invalid credentials (49) [15:52] zul: d'oh! when i just install libvirt-bin, the perms are right [15:52] hallyn: so its an upgrade issue then [15:54] zul: no, [15:54] zul: i think libvirt is doing it itself [15:54] armaan, hey there [15:55] adam_g, ping [16:01] New bug: #1025336 in php5 (main) "installing php5-curl reloads apache *before* it's set up" [Undecided,New] https://launchpad.net/bugs/1025336 [16:03] koolheadd17: hey... === Lcawte|Away is now known as Lcawte [16:06] zul: steps so far to reproduce. spin a new instance. install libvirt-bin. ls -l /var/lib/libvirt -> images looks good. now connect with virt-manager as a user. images dir is no longer good [16:06] armaan, the document did not help you to configure ldap [16:06] ? [16:06] checking to see if it's libvirt itself, or virt-manager [16:06] koolheadd17: Nope [16:06] koolheadd17: got the same error [16:06] armaan, can you do one thing [16:06] koolheadd17: ?? [16:06] setup a fresh instance [16:06] and try everything from scratch on that [16:07] if u get same error [16:07] koolheadd17: already did it last night. [16:07] hallyn: maybe there is something in virt-manager that needs to updating [16:07] dont forget to file a bug [16:07] armaan, file a bug dude [16:07] zul: could be. [16:08] koolheadd17: i think i have to do it now. [16:08] hallyn: erm - I have a setup like that [16:08] koolheadd17: do you know someone who might have worked with ldap [16:08] RoyK: this only seems to affect libvirt on quantal [16:09] RoyK: results in non-root users being unable to remotely spin up images (unless they use ~/VirtualMachines for storage) [16:09] quantal? [16:11] oh [16:11] hallyn: I usually don't run pre-alpha software on my production machines ;) [16:12] armaan, i was able to get ldap configured without any issue in oneiric [16:12] so am saying [16:12] file a bug [16:13] RoyK: we're here to make sure it works for you when you switch :) [16:14] hallyn: good point ;) [16:14] * RoyK hardly uses anything but LTS on servers, though [16:17] koolheadd17: okie... i will try once more on a fresh vm if it failed i will file a bug. [16:22] yes sure [16:23] hi guys, we are working on updating the zentyal packages for 12.10, but this is blocking us, any openldap maintainer around that can have a look? https://bugs.launchpad.net/ubuntu/+source/openldap/+bug/82853 [16:23] Launchpad bug 82853 in openldap "Add support for the smbk5pwd overlay" [Wishlist,Confirmed] [16:26] zul: ok, yeah, virt-manager defines a storage pool, and it ends up with '0700' in the xml for it [16:27] hallyn: ah hah thought so [16:27] zul: i assume libvirt is filling that in as a default. not sure where we want to fix it [16:27] zul: gotta run a for bit, bbl === matsubara-afk is now known as matsubara [17:15] zul: all right. so we could use commit 8a544719aa2fe92332756ea0483844b81d4d56ae which should change the default from 0700 to 0755. but that's still not quite right :) [17:16] zul: so we may just want to custom patch it ot make 0711 the default. not sure [17:16] hallyn: up to you === Aaton_off is now known as Aaton === Hetep is now known as Asar === cpg|away is now known as cpg [18:05] rbasak, are you around? i was looking at zul's patch at http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/quantal/euca2ools/quantal/view/head:/debian/patches/add-armhf-support.patch [18:05] which is now upstream at https://github.com/eucalyptus/euca2ools/commit/b89af8fa79d97a94c033a3d97610035b723f448c [18:12] New bug: #1006553 in juju "local provider machine agent uses 100% CPU after host reboot" [High,In progress] https://launchpad.net/bugs/1006553 [18:12] * hallyn goes to check if that's his fault [18:13] not me [18:36] zul: mdeslaur: any objection to http://people.canonical.com/~serge/libvirt-perms.debdiff ? [18:36] hallyn: nope === cpg is now known as cpg|away [18:40] zul: ok, thanks, will push. (not *convinced* it's waht we want in the end, is all) [18:40] hallyn: I, uh...not really, if it doesn't break anything [18:40] hallyn: does virt-manager still work if you do that? [18:41] mdeslaur: without the upstream patch virt-manager is broken. With just the upstream patch, virt-manager will cause /var/lib/libvirt/images to be 0755 (adding group and world read perms) [18:41] mdeslaur: yes. [18:41] mdeslaur: with the upstream patch and the additional one-liner, /var/lib/libvirt/images is back to 0711 as it used to be. [18:41] the q is, [18:41] will it confuse other people when they make a new custom storage pool [18:42] well i'll push this for now. people can always jsut add the permission tag to their xml if it bothers them === cpg|away is now known as cpg [18:42] hallyn: ok [18:42] mdeslaur: thanks [18:47] Daviey, ping === remix_auei is now known as remix_tj [19:19] wehre can i find the shutdown logs from last shutdown? [19:19] nfsen is hanging up on shutdown so i want to see if there's any more info [19:19] utlemming: ping [19:20] it's also hanging on startup [19:31] stgraber: so the reason i hadn't done reboot yet is that lxc_start goes the lazy route - it doesn't clean anything up, just re-execs itself with origina larguments [19:31] New bug: #1025418 in openssh (main) "Using ProxyCommand w/a non-existant host results in infinite spawns." [Undecided,New] https://launchpad.net/bugs/1025418 [19:31] irv: should be in dmesg [19:31] i'm going to see just how much it would take to clean up [19:31] that's the current, though [19:32] irv: /var/log/syslog or /var/log/syslog.1 for last one [19:49] stgraber: yay, the only thing *crucially* needed was closing the /var/lib/lxc/container/command socket. mind you i'm sure this'll leak memory like a seive for now, but it works [19:55] koolhead17 keeps missing me :( [19:59] stgraber: pushed my fix, reboot should now work. [19:59] hallyn: cool, thanks. Will test later (still need to finish adding get_config_item/clear_config_item) [20:00] stgraber: hopefully i didnt' mess you up by updating the patch [20:00] oh, your changes aren't to that patch :) [20:01] hallyn: yeah, my changes are in debian/python-lxc, you're the only one touching that patch :) [20:52] lynxman: as the common-law maintainer of ipxe, any comments on bug 1025239 ? [20:52] Launchpad bug 1025239 in ipxe "kvm-ipxe does not support https" [Medium,Confirmed] https://launchpad.net/bugs/1025239 === Pupeno_W_ is now known as Pupeno_W [21:16] I have a NFS question. In NFS v4. How would I use the UID/GID of the client (passwd/group) to map. I do not even need the server to map user/group .. I've been using NFS v3 for misc file servers and the server has never mapped the UID/GID .. [22:06] zul: could you take a look at bug 82853 [22:06] Launchpad bug 82853 in openldap "Add support for the smbk5pwd overlay" [Wishlist,Confirmed] https://launchpad.net/bugs/82853 [22:07] zul: AFAICT the reason not to have smbk5pwd was that we weren't building openldap with heimdal-dev, but nowadays we are, so the delta doesn't make much sense and is preventing some folks like bencer from using the tool [22:07] we need that for zentyal-users, having slapd synced with samba4 [22:07] stgraber: im not comfortable of adding this for an SRU [22:08] zul: unless there's some other required build-dependecy that'd need MIRing to enable that, I'd suggest reverting that part of our delta and having that package built and put into universe (so we don't need to MIR extra runtime dependency) [22:08] zul: never said I want that in an SRU [22:08] and i dont really maintain openldap either [22:08] zul: for 12.10 [22:08] https://launchpad.net/~zentyal/+archive/2.3/+files/openldap_2.4.28-1.1ubuntu4_2.4.28%2B51~precise1.diff.gz [22:08] ok [22:08] didnt check that debdiff [22:08] zul: bencer is working on samba4 support in zentyal for 12.10 and it's apparently a requirement [22:08] but i was told just enable that on debian/rules and the debian/control stuff [22:09] stgraber: well if someone wants to drive that than im fine with it [22:09] Hi, does anyone encounter problem setuping cLVM. When i run vgcreate --clustered y mygrp /dev/sdb it reports it is created succesfully but when i try vgdisplay it says "Skipping clustered volume group..." and lvcreate does the same. am i doing something wrong? [22:10] bencer: can you give me a debdiff against quantal's current openldap? I can then review and sponsor if that looks good [22:10] stgraber: deal === cpg is now known as cpg|away === cpg|away is now known as cpg === Aaton is now known as Aaton_off [23:05] <\sh> anyone familiar with freeipa-client and authentication issues? === Lcawte is now known as Lcawte|Away === n0ts_off is now known as n0ts [23:27] hi all [23:34] jcastro, where are you ppl? [23:41] koolhead11: pateince [23:41] we're not around 24/7 :P [23:47] TheLordOfTime, :P