[00:32] omg ipv6 is so hard [00:41] Hello. I'm trying to set up a test MAAS/Juju/openstack environment. I've got every node running, and according to juju status they are all setup corrently as far as I can tell. The big issue is that I cannot log into the Openstack Dashboard using admin and the admin-password provided in the keystone config file when deploying keystone. Any ideas? === pug is now known as Guest11967 === Guest11967 is now known as DaveR === acidflash_ is now known as acidflash [04:17] how would I auto start an openvpn client? I can connect using sudo openvpn client.ovpn but not with service openvpn start === cpg is now known as cpg|away === ibiris|afk is now known as ibiris === cpg|away is now known as cpg === ibiris is now known as ibiris|afk === ibiris|afk is now known as ibiris [08:06] I have a question about software raid on server 12.04 [08:06] I have ext4 with journaling on top of raid1 using mdadm [08:07] when the array re-syncs, the ext4 journaler hangs for LONG periods of time and locks up the kernel [08:08] has anyone else experienced this? any ideas? it's a serious problem. [08:12] I'm seeing bug reports on the issue from 2010, but it seems like the devs could never track it down [08:26] New bug: #1091114 in checksecurity (main) "fuse mount cause problem to find -xdev" [Undecided,New] https://launchpad.net/bugs/1091114 [08:37] New bug: #1089711 in nova (main) "nova.conf unreadable is hard to diagnose" [Undecided,New] https://launchpad.net/bugs/1089711 [08:59] hey, guys short question, how can i see the reason for an security update via console? [09:10] TheLordOfTime: yes. [09:31] jstephan: you could do something like "apt-get changelog /precise-security" [09:31] ah, okay, thx === alaing_ is now known as alaing === alaing_ is now known as alaing === cpg is now known as cpg|away === alaing_ is now known as alaing [10:58] hi all, got a issue with mdadm, taken the disks from one server installed them in another server, but they will not reassemble. [10:58] how can I reassemble them [11:01] the homehost will not match, so you either assemble them by hand, override their hostname to match, or make it ignore homehost setting. === asac is now known as asac_the_hacker === mcclurmc_away is now known as mcclurmc === asac_the_hacker is now known as asac === alaing_ is now known as alaing [14:26] hi. [14:26] just tried to install php5-mysqlnd this would remove phpmyadmin [14:26] as far as i read this is considered to be a bug [14:27] because phpmyadmin depends on php5-mysql [14:28] phpmyadmin depends on: php5-mysql | php5-mysqli | php5-mysqlnd [14:28] So it should accept any one of those [14:29] Jever: ^^ [14:29] i'l [14:29] i'll try [14:30] doesn't work [14:30] if i install php5-mysqlnd it removes phpmyadmin and php5-mysql [14:31] if i try to install phpmyadmin afterwards it will remove php5-mysqlnd again and install php5-mysql [14:31] Try requesting both at once. Hopefully it'll tell you why it can't fulfill thta [14:32] Ah [14:32] Looks like php5-mysqlnd was added as an option after precise. It's present in raring. [14:32] nope can't install both [14:32] so what does that mean for me [14:34] Assuming that the packages will work together, it sounds like an acceptable bug as you described. I'm not sure if a dependency change would be acceptable for a stable release update though. [14:35] https://bugs.launchpad.net/ubuntu/+source/phpmyadmin/+bug/1012670 [14:35] Launchpad bug 1012670 in phpmyadmin "Dependency on php5-mysqli" [Undecided,Confirmed] [14:35] So a bug would be marked as fix released as it's fixed in the current development version, and somebody would have to ack it for an update to 12.04 if it is acceptable (I'm not sure if it is) [14:35] this was 6 months ago [14:35] are we sure phpmyadmin works with mysqlnd? [14:35] okay but is there a quick fix for me to install it anyway? [14:35] In the meantime, you could override the dependency, or use a more recent release [14:36] i would like to override it to test if it works for me [14:36] patdk-wk: I'm not sure, so this is all assuming that it is. However the latest php5 package in raring seems to allow it. [14:37] Looks like it works in 12.10. [14:38] can't update though. any way to ignore dependencies without using dpkg manually? [14:38] I think you have to use dpkg manually [14:39] which means that i won't get updates? [14:39] Easiest hack would be to install both using apt-get, then you'll have the right package in /var/cache/apt/archives to install with an override using dpkg [14:39] Things will break if you use updates [14:39] hmpf. i'll stay with php5-mysql [14:39] thanks for your help [14:41] You could rebuild phpmyadmin in a PPA if you're feeling adventurous [14:41] Sorry I can't give you a better option! Thanks for bringing it up and finding the bug. I've updated the bug. [14:43] no problem. just wanted to test it and found out that it doesn't work yet [14:44] rbasak, same issue for php-mdb2-driver-mysql [14:45] guess a bunch of packages need to be updated [14:45] Yes. Or the different drivers should all provide a virtual package and consumers like phpmyadmin could just depend on that [14:45] hell, I would likely opt to maky php5-mysql a meta package that installs php5-mysqlnd by default [14:45] How many are there? [14:46] 3 [14:46] mysql mysqli mysqlnd, though mysql and mysqlnd are *compatable* [14:46] currently atleast [14:48] php-mdb2-driver-mysql doesn't seem to conflict with anything though, so phpmyadmin can be installed with it [14:48] I don't know whether it would use it or not [14:48] hmm? [14:48] You said same issue with php-mdb2-driver-mysql? [14:48] yes [14:48] it depends on php5-mysql [14:48] Oh, I see. [14:48] install php5-mysqlnd wipes it :) [14:49] oh, it's universe [14:49] So is phpmyadmin [14:50] guess I can just use pear instead for that [14:52] The rdepends list for php5-mysql is quite large [15:06] New bug: #1000175 in openssh (main) "Unable to copy id to a server that is listening on non-default port (dup-of: 518883)" [Wishlist,Confirmed] https://launchpad.net/bugs/1000175 [15:31] smoser: ahem. Oops. Fixed. I did wonder why the cloud-init maintainers team was going to review it! [15:33] smoser: btw, I couldn't figure out what was going on with the packaging, so I left it. debian/control exists in both trunk and the packaging branch for example, and are different. [15:34] rbasak, its just an upstream that has its own debian/ [15:50] hi folks [15:50] any maas peeps around? [15:52] …tumbleweed...... [15:53] rbasak, did you know that cloud-init probalby as of precise will output the full fingerprint to the console ? [15:53] er... the full key [15:53] so you dont have to do the fingerprint and ssh-keygen check. [15:54] * rbasak looks [15:54] smoser: I see. That's useful! [15:54] rnbrady: maybe try the #maas channel [15:55] smoser: still need to support the old method though, do you think? Or forget about it if precise supports it? [15:55] i think it'd be nice to support the old method. [15:55] pmatulis: ah yes, thanks! [15:55] magically fall back to it if only that form is found. [15:56] OK. I'll have to rethink the CLI a bit though [15:56] I guess I need to make it grep-fingerprint and grep-public-key, import-fingerprint and import-public-key, and fix will dtrt. [15:58] that is https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/893400 [15:58] Launchpad bug 893400 in cloud-init "cloud-init: Output public ssh host key (for known_hosts)" [Low,Fix released] [15:58] just for reference [15:58] Thanks! [16:36] I've installed MAAS. I only have one server (acting as master) now that I wish to install openstack. In about two weeks I'm adding another node. Am I suppose to add my master server as a node to maas? [16:43] I've installed MAAS. I only have one server (acting as master) now that I wish to install openstack. In about two weeks I'm adding another node. Am I suppose to add my master server as a node to maas? [16:44] !patience [16:44] Don't feel ignored and repeat your question quickly; if nobody knows your answer, nobody will answer you. While you wait, try searching https://help.ubuntu.com or http://ubuntuforums.org or http://askubuntu.com/ [16:44] savr, you can askin #maas, fwiw. [16:45] hello [16:45] savr, the maas server is not a node in maas. is that what you were asking ? [16:45] hi all. my brother has a RAID-5 on three 2TB drives and want to add another three drives to make it a 6-drive RAID-6. Anyone here that knows if this can this be done in a single operation, or will he have to add a single drive at a time? also, he's currently on Lucid, kernel 2.6.32 and IIRC that doesn't support migrating to RAID-6, is this right? I guess an upgrade to Ubuntu Precise shouldn't be too hard... [16:45] smoser: currently maas is telling me I have 0 nodes [16:45] right. [16:45] you have 0 nodes. [16:45] am I suppose to add the master as a node? [16:46] you really ahve to add other nodes. [16:46] the maas master is not a node in itself. [16:46] (if it was, and it could manage, that would then force itself to re-install itself) [16:46] I shouldn't really be using maas then? [16:47] maas allows you to treat a group of hardware as provisionable like a cloud (which juju uses to install openstack). juju install of openstack is several nodes (glance, compute, swift...) [16:47] so with just 1 node, maas is not terribly useful. [16:47] I don't really know why I installed maas. I'm following the guide to use juju to install openstack [16:48] savr, for a single "all in one openstack", the best solution that i'm familiar with is devstack. [16:48] the reason why I'm installing openstack is because I intend to have a few servers and want the ease of being able to move services around [16:48] but that wont allow ubut that will not give you any real path to grow [16:49] for some reason I think it is a billion times easier to do all this manually [16:49] as that is what I'm use to and all these new tools seem a lot more complicated than editing a few config files [16:49] Perhaps one needs to suffer the pain of doing all this manually before one accepts the usefulness of having automated tools for all of it :) [16:50] rbasak: I've done all this manually many times [16:50] For a single server, it's probably not worth it [16:50] I know how to do it manually [16:50] What you have is a server, not a cloud! [16:51] I'm going to have 4 servers and maybe 5 services running inbetween them all [16:51] I want them to failover and load balance [16:51] to set this up manually is like 1 hour work for me [16:51] Sounds like that's your best bet then [16:51] to understand maas juju openstack cloud foundry etc is much more difficult [16:52] ok then [16:52] time to nuke this maas install [16:52] does juju make any sense for me? [16:52] or cloud foundry or open stack? [16:54] savr: if you intend to have a smallish static environment, i don't think so [16:54] i would say that if you're comfortable managing virtual systems in some other way, 4 servers and 5 services is a small number and you will incur significant overhead in openstack. but having a system like openstack in place will help you to grow. [16:54] thanks! [16:55] should I be using vps? [16:55] If you can do everything you want in one hour's admin time, then I don't think there's any other solution that makes sense for you. But when you find yourself trying to do something that will take many hours or days of admin time, or you find yourself writing tools for these tasks, or you need to study services in detail to learn how to use them, then come back and take a look at juju :) [16:55] juju would be useful to you as it would enable you to provision codify your installation of your services. [16:55] but juju needs a provider, and there is only the "local provider" which uses lxc that would really help you. [16:56] juju is a massive learning curve for me right now [16:56] I tried using the local provider [16:56] full of issues [17:00] the local provider is for like your laptop, you wouldn't run it on your servers [17:01] so does it make sense to virtualize [17:01] one of the services is wordpress [17:01] another is virtual desktop [17:02] then some in house web software we use [17:02] etc [17:05] jamespage: Sigh, powerpc & armhf are so slow [17:06] * TheLordOfTime was whining about powerpc last week when it took 3 days for a package sync to actually reach raring and get out of raring-proposed. [17:06] powerpc's... not fun. [17:07] rbasak: smoser: would you virtualize in this instance? [17:08] i'd probably lean towards virtualization and some configuration management solution [17:09] If you can reinstall it all in an hour then I don't think it matters what you do. Just reinstall when you want to move something. [17:11] I'm with a great provider so reinstalls are super ease. normally takes 5 minutes. Installed with ssh keys and can auto run post install bash script. [17:12] smoser: like what software? [17:12] * maswan is using ganeti to handle a bunch of service VMs on a few hosts [17:13] puppet and chef are configuration management system.s [17:28] CVE-2012-5642 is awesome. [17:28] Daviey: ** RESERVED ** This candidate has been reserved by an organization or individual that will use it when announcing a new security problem. When the candidate has been publicized, the details for this candidate will be provided. (http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2012-5642) [17:37] New bug: #1080841 in cloud-init "should reload configuration if an upstart job is added" [Medium,Fix committed] https://launchpad.net/bugs/1080841 [17:39] mysql keeps disappearing on my 12.04 ubuntu server. i see no record of either an oomkill or a segfault in /var/log/syslog any ideas what else to look for to see why it keeps disappearing? [17:40] maco: is there anything in the mysql logs? [17:41] sarnold: /var/log/mysql.err is blank and /var/log/mysql/ is empty and /var/log/mysql.log is a 0byte file [17:41] maco: very curious. ) [17:41] maco: does df show you have adequate storage space available? [17:42] sarnold: bout 15gb free [17:42] zless /var/log/mysql.log.1.gz says it's a binary file, 20bytes [17:42] nothing legible in there [17:43] as a stopgap i modified the /etc/init/mysql.conf to retry 10 times instead of only 2, over the course of 90 seconds instead of 10, which has slowed down the rate at which i suddenly discover my sites are offline, but... [17:43] is it too-open? MySQL seems to be a walking pile of CVE generators, maybe someone is having fun with you.. [17:43] last i checked, PHP's more evil than that, sarnold. [17:44] TheLordOfTime: no argument there, php is also a giant pile. [17:44] and wordpress even more so ;) all of which are on my server, but i have htaccess-locked the wp admin pages for safety there [17:44] don't get me started on other stuff in Universe... [17:45] dont think universe has wp updates fast enough [17:45] all of the mysql.log.[0-9].gz's are 20bytes [17:46] I think 20 bytes is the usual size of zero-byte-files that gzip compresses... [17:46] ah [17:47] as to "too-open" the my.cnf is set to only listen on localhost (that's the default) so outsiders shouldnt be talking to the process [17:48] hah im amused that in addition to TheLordOfTime there's also TheDrums [17:48] i think that was spawned by my taking this nick :p [17:49] thanks all! [17:51] oooh wait it is oom. shouldve done a case-insensitive grep [17:51] hrmph. time to tune apache so it stops eating half the ram [17:52] * TheLordOfTime blurts out "NGINX RULZ" then facedesks hard for being biased. === mcclurmc is now known as mcclurmc_away [19:00] hello world! [19:00] hi proxmox or ganeti? or makes no difference/ [19:04] think I'm going with proxmox as it can be installed by the datacentre [19:09] savr, KVM? [19:09] yep [19:10] I've not looked at ganeti for a few months, but it didn't make sense for us here [19:10] I ran proxmox 2.1/2.2 for awhile on a test cluster, but it kept forgetting what my machines were named [19:10] tgm4883: what are you using? [19:11] savr, KVM [19:11] well, kvm with libvirt I suppose [19:11] no management software? [19:11] on our production cluster we run ESX [19:12] savr, nope, we just run these 4 servers when we need to setup some test stuff [19:12] nothing that we are super worried about [19:12] MaaS and Ubuntu Cloud don't make sense for us either it seems === ibiris is now known as ibiris|afk === yofel_ is now known as yofel [20:42] I have a question relating to kerberized nfs: how would I enable a service (say apache) to access a kerberized nfs mount (e.g. a users home/public_html)? is there a "standard" solution to this problem? [20:46] New bug: #811417 in apache2 (main) "Apache httpd local DOS when serving pipes due to TOCTOU" [Low,Incomplete] https://launchpad.net/bugs/811417 [20:49] Good evening. We just got hit by some weird attack/exploit/something which brought our machine to it's knees. Somehow, it managed to hold up and I got this screenshot before it went down: http://puu.sh/1C3Sb/e7bb806225aedbd720508e27d1261d6d [20:49] Does anyone have any idea what this is? [20:49] It's also spamming the syslog with out of memory, too many open files [20:50] etc [20:50] it doesn't need to be a DoS - it can just be I/O hanging [20:50] but the load is pretty high :P [20:50] it happens on two machines [20:51] at the exact same time [20:51] (couple seconds in between) [20:51] sounds like a ddos [20:51] yeah, but it's not a BW related one [20:51] the network traffic is normal [20:51] as far as we know anyway [20:51] tried to sniff it with tshark/tcpdump? [20:52] it goes down, so we're unable to [20:52] I was going to attempt using a kvm to login, but it just crashed when i tried :/ [20:52] 10-20k load is pretty amazingly ugly :( [20:54] mikeey: try enabling netconsole - you might get some output from that that may be drowned because of memory shortcomings [21:36] New bug: #1091430 in qemu-kvm (main) "9p fails with readonly+non-root due to O_NOATIME" [Undecided,New] https://launchpad.net/bugs/1091430 [22:17] Sounds awesome, http://bugs.debian.org/696195