[00:19] hlan: euca-get-console-output can be helpful [00:19] hlan: or ec2-get-console-output if you prefer :) [00:24] SpamapS: is this the same as "Get system log" in the ec2 gui? [00:24] hlan: probably, I've never used the gui === fysics_ is now known as fysics === Arc_ is now known as a5m0 [01:10] I just installed the daily precise ami I found here http://uec-images.ubuntu.com/precise/current/ [01:10] as I probably need to update the kernel ASAP. [01:10] hey there [01:10] I see some weird daemons running... "whoopsie" [01:11] got an issue with 3 dedicateds when we set a keepalive on the ssh connection is constantly refused [01:11] ending up in us having to reimage the servers ¬.¬ [01:11] the biggest problem is that dnsmasq is running for some reason and it's preventing me from booting my DNS server [01:14] how can I remove dnsmasq... apt-get tells me that dnsmasq isn't even installed?? [01:14] and I can't find anything in init.d [01:17] i had a do-package-upgrade going and i got disconnected from ssh… can i resume the upgrade somehow (or the ssh session) ? [01:18] release-upgrade* [01:25] i'm having problems with ssh aswell [01:27] blendedbychris: I know it won't help this time, but you may want to look at using screen or tmux in the future... [01:31] I'm attempting to setup a MaaS server on Ubuntu 12.04, but my nodes keep hanging at "init: cloud-init-nonet main process (256) killed by TERM signal" [01:31] Can anyone help? [01:31] I can try [01:31] Thanks :-) [01:32] having said that I am more familiar with the webapp side of things than cloud-init [01:32] smoser, any idea? ^ [01:33] hmm, if I could login to the nodes themselves or access their logs that would really help [01:36] New bug: #984499 in net-snmp (main) "snmpd fails to upgrade on do-release-upgrade -d from oneiric to precise" [Undecided,New] https://launchpad.net/bugs/984499 [01:41] after upgrading to precise htop shows weird ââ characters in the process list... weird [01:41] perhaps some unicode support or something that messes with the layout? [01:42] I think I may have found the problem... [01:43] http://pastebin.com/JPw9F5FN [01:43] My apache error log is full of these and they appear whenever the cloud-init-nonet runs [01:44] any ideas? [01:44] crap, just tested upgrading to ubuntu precise to fix my ec2 kernel crash, just had the exact same behavior :/ [01:45] so I just ruled out upgrading from 3.0 to 3.2 [01:45] and the system log is dead silent as usual... [01:45] DiabolicalGamer: ah I know [01:45] DiabolicalGamer: the clock is wrong on the node [01:45] lol [01:45] is that all? [01:45] well either the node or the maas server [01:46] *facepalm* [01:46] yeah, it breaks OAuth if they are too different [01:46] it must be the nodes then because I configured the system clock when I installed ubuntu on the cloud controller === jtv1 is now known as jtv [02:02] is there a way to force the nodes to run ntp-update? [02:03] *ntpdate [02:12] OMG it worked! [02:12] Thanks bigjools [02:12] DiabolicalGamer: yay! [02:13] we know this is a bug, just need to work out a good way of fixing it [02:13] That error should be much more noisy. [02:13] like "Hey guy you forgot to set your clock!" [02:13] lol [02:13] it should not be noisy at all, it should work [02:14] I had to set my clock to a weird time to get it working [02:14] weird as in "eleventy past fiveses" ? [02:15] I think it has to do with timezones or something [02:16] I just did the math and changed it to what "should" work and it did [02:18] *punches sshd_config in the face* [02:23] testing? [02:24] Hey all.. I just upgraded to ubuntu precise and my interfaces are not coming up for some reason… any help? [02:24] They are still configured from my lucid install [02:24] you updated to 12.04 beta? [02:24] yes [02:25] what's in the interfaces file [02:26] eh crap uhm [02:26] "/etc/network/interfaces [02:26] ignore the " [02:26] gotta type this out really stupid pptp is all forwarded [02:26] auto eth0 [02:26] iface eth0 inet static [02:26] pre-up /usr/sbin/ethtool -s eth0 speed 1000 duplex full autoneg on [02:27] use paste.ubuntu.com mate [02:27] address 10.16.26.135 [02:27] can't :( [02:27] :'( [02:27] I promise I would [02:27] I'd have to disconnect from my kvm [02:27] netmask 255.255.255.192 [02:27] is 10.16.26.135 an internal? [02:27] up route add -net 10.0.0.0/8 gw 10.16.26.129 [02:28] yes [02:28] really just worred about that atm [02:28] i can't see an external ip [02:28] that why you can't connect? [02:28] i'm just worried about the internal atm… ifconfig has neither interface listed just lo [02:28] hmm [02:29] maybe a driver issue? [02:29] or pre-up ? [02:29] does iftop bring any interfaces up [02:29] oh ya… there's no such thing as /usr/sbin/ethtool [02:29] was this a dist-upgrade from your provider or using one of their images? [02:30] sorry brb one sec [02:30] well provider to precise [02:30] i'm going to comment out the pre-p crap [02:31] ya that fixes it [02:31] any idea if that's necessary? [02:33] pass varies by provider [02:33] tbh on the dist upgrade [02:33] it should have asked you about changes to the interfaces file [02:34] showing you what lines it wanted to add and which lines it wanted to remove. [02:34] well I'm sure there were no changes [02:34] *sets up 4th dedicated server in 24 hours* [02:34] well technically 7th if you include the 3 i bricked ¬.¬ [02:34] but /usr/sbin changed to /sbin [02:37] on my fresh 11.10 installs i have one in both /sbin and /usr/sbin [02:39] I don't have any "/proc/sys/fs/inode-max", why is that? :/ [02:39] nor do it [02:39] i* [02:46] 3 terminals open one on each desktop pasting the same line into every one. Joy. lol. [02:47] is there an easy way to create an image of a server? [02:48] clonezilla [02:49] Laice: does that work for ext4? [02:49] or a better question why would i use that vs dd [02:53] bleh [02:54] pass [02:54] huh? [02:54] i'm passing the question :P might say on clonezilla.org [02:54] i don't even know how to get started using clonezilla heh [02:54] especially in a hosted environment [02:55] if all else should fail PUSH ALL OF THE BUTTONS. something will happen eventually [02:55] he [02:55] the faster you push the buttons the better the server admin you are :p [02:56] plus it makes you look more clever. Mostly as they can't see you're flailing around xD [02:56] bleh this isn't worth the effort [02:57] how much do you pay for your vps? [02:57] i wanted to clone a clean precise install so i didn't have to go through that networking crapola [02:57] oh we use softlayer and get ripped of royally for our vps [02:57] get a kimsufi dedicated [02:57] cheap as hell [02:57] not my decision [02:57] is in europe anyway [02:57] heh [02:57] money speaks louder than words :P [02:58] i use linode for my nonwork stuff [02:58] the big factor though for work is we need some sort of mass storage [02:58] and most vps providers don't really offer that [02:58] it amazes me that people go for a VPS when dedicateds work out to be better value and in many cases cheaper. [02:59] yea when you start adding more storage dedicated is cheaper [02:59] 20$ per month gets me 5TB bandwith a 4 core 1.8ghz atom and 2gb ram. perfect for small hostings [02:59] which is why we have a cluser of them :P [02:59] atom wtf? [02:59] cluster* [03:00] better than someone screwing your vps's cpu [03:00] linode is the same but you share 8cpu 4cores [03:00] euch [03:00] it's the "share" bit i hate [03:00] lmfa my news just had a image they showed online that had a watermark "Web Viewing Only" [03:01] er [03:01] on tv rather [03:01] *picards* [03:01] Laice: do you know of a quick way to copy a server to another? [03:01] same hardware and drives [03:01] clonezilla. [03:02] would i boot the livecd on a new server? [03:02] the* [03:02] does support ext4 looking at this http://clonezilla.org/ [03:04] looks like you can save the image to another partition [03:04] my issue is i use the whole drive though? [03:05] bleh [03:05] too much of a hassel [03:05] no pain no gain [03:05] or in this case, no pain in reimaging :P [03:06] i swear i'm gonna make a script for configuring the server automatically once it's reimaged [03:06] the stupd little 1gb storage they give you for this kind of crap isn't even working [03:06] lol made me laugh they offer you at ovh for $1 per month a technitian will walk into the data cente and plug a usb stick in. [03:07] bargain [03:07] probably would get annoyed about me ringing up and saying "uh, could you plug it into this server now please, now this one, now this one...." [03:07] retards just jumped into the pool outside…. it's like 60degs [03:08] "ohhh it's so cold" [03:08] anyone here knows how to assign variables in #cloud-config user-data so I use the values multiple times? [03:15] blendedbychris: 60 is warm unless you're a silly yankee imperialist [03:15] Or I guess a scientist... [03:17] what's that in real money (*C) [03:17] Okay another road block here... MaaS is working, but now JuJu is complaining. [03:17] or live in texas [03:20] http://pastebin.com/wTWBAXeS [03:20] Any help would be greatly appreciated. [03:21] This whole cloud thing is actually secrets just a game to make up as many acronyms as possible, isn't it [03:21] s/secrets/secretly/ [03:22] I was beginning to wonder that myself [03:29] http://pastebin.com/LqkjUzvR [03:29] So it seems like juju is breaking the maas api [03:29] Any ideas? [03:32] http://www.youtube.com/watch?feature=player_embedded&v=DbkvYMIjIu4 === jtv is now known as jtv-afk [03:43] DiabolicalGamer: looking for you now [03:43] thanks [03:44] DiabolicalGamer: is this the first time you ran bootstrap? [03:44] yes [03:44] which version of juju are you using? [03:45] if it's the latest from the ppa that's ok [03:45] latest ppa [03:45] the whole maas server log would be useful too [03:47] the /var/log/maas/maas.log? [03:48] DiabolicalGamer: yes [03:52] bigjools: Here -> http://pastebin.com/i9VPmJAn [03:54] DiabolicalGamer: you have done something weird with your database [03:54] It's a fresh install [03:55] you added some nodes? [03:55] Yes [03:55] Two [03:55] what status are they [03:55] ? [03:55] Ready [03:55] ok so you commissioned them already? [03:55] Yes [03:56] can you try destroying the juju env and bootstrap again [03:56] this is obviously a bug but I want to try and work out where the problem is before you delete the evidence :) [03:57] hmm I'm not sure what you mean by destroying juju [03:57] do u mean purging? [03:57] juju destroy-environment [03:57] oh lol [03:58] something is out of sync between juju and maas [03:58] so let's start from scratch [03:58] okay done [03:59] and bootstrap again [03:59] hmm same error [04:00] ok [04:00] 2012-04-17 20:59:48,751 INFO Bootstrapping environment 'maas' (origin: ppa type: maas)... Unexpected Error interacting with provider: 500 INTERNAL SERVER ERROR 2012-04-17 20:59:49,100 ERROR Unexpected Error interacting with provider: 500 INTERNAL SERVER ERROR [04:00] not sure if this works on the packaged version, but try "maas dbshell" [04:00] should get a psql prompt [04:01] lol sudo thinks maas is a user not a command [04:01] it is :) [04:02] I know, but it's asking for the password linked to the maas account rather than mine when I sudo [04:03] I don't know it's password [04:03] it doesn't have one IIRC [04:04] sudo -u maas maas dbshell ? === jtv-afk is now known as jtv [04:06] psql: fe_sendauth: no password supplied [04:06] won't let me in [04:06] even if I sudo su - [04:06] darn it, ok sorry I'm not too familiar with what happened in the packaging, I only worked in the dev environment. Ummm let me think [04:07] try adding a PGUSER=maas [04:08] * bigjools clutching at straws [04:09] hmm no go [04:09] even used passwd to assign maas a password [04:09] still nothing [04:09] yeah it's the DB password [04:09] DiabolicalGamer: did you upgrade or where did you see that issue? [04:10] roaksoax, my man! [04:10] DiabolicalGamer: the password one? what versions of maas are you using? [04:10] bigjools: lol [04:10] one sec [04:10] roaksoax: trying to get a psql prompt on his maas db [04:11] so I can see what state the nodes are in etc [04:11] DiabolicalGamer: do you have a ssh key created? [04:11] yes [04:11] fjlacoste: he's not getting that far [04:11] DiabolicalGamer: the pastebin you provided is cropped [04:11] following: https://help.ubuntu.com/community/UbuntuCloudInfrastructure [04:11] something is blowing up with bootstrap [04:12] DiabolicalGamer: can you paste them again with full line length? [04:12] (without the $ add the end) [04:12] and from the logs, some of the nodes are already allocated but in status READY [04:12] fjlacoste: http://pastebin.com/i9VPmJAn [04:12] sure moving the log from linux to windows must've messed it up [04:13] the full one you did is fine [04:13] sudo maas dbshell needs a password [04:13] bigjools: that error is a red herring [04:13] it's because public-keys raise a 404 [04:13] you'll find the DB pass in /etc/maas/maas_local_settings.py [04:13] which is expected behavior [04:13] fjlacoste: not convinced about that [04:13] bigjools: i am *sure* about this [04:13] I didn't have my ssh key set at first, then I set it. [04:13] i have seen that error several times [04:13] fjlacoste: ok :) [04:13] and it's benign [04:14] simply add ssh keys to your account and it will disappear [04:14] but it's not the root cause [04:14] weird [04:14] wtf is causing that [04:14] the maas api doesn't care about ssh keys [04:14] I already have an ssh key added... [04:14] maas api for juju I mean [04:15] okay I got into the postgre shell [04:15] ah, i know what it is [04:15] that's because the owner isn't created [04:16] fjlacoste: I think commissioning left the user allocated on the node [04:16] yeah [04:16] we fixed that [04:16] we did [04:16] but which version is he using? :) [04:16] but that's not in archives [04:16] DiabolicalGamer: are you running the latest package from testing? [04:16] DiabolicalGamer: but do not upgrade to the testing ppa for now [04:17] the new maas-provision package is broken [04:17] on upgrade [04:17] No all stable [04:17] yeah [04:17] so that one has that bug [04:17] yeah [04:17] you can workaround by resetting the owner [04:17] in the DB [04:17] or you can wait a few minutes [04:17] for roaksoax to upload a fixed maas-provision package [04:17] and then update to the testing ppa [04:18] let me know if you want the DB surgery workaround [04:18] it's actually documented in the bug [04:18] let me paste you the bug number [04:18] DiabolicalGamer: in the dbshell, update massserver_node set owner=null; [04:18] can I manually download the necessary deb package? [04:18] DiabolicalGamer: no, because it depends on maas-provisioning [04:19] DiabolicalGamer: https://bugs.launchpad.net/maas/+bug/981068 [04:19] Launchpad bug 981068 in maas "AssertionError when attempting to acquire a node after commissioning" [Critical,Fix released] [04:20] bigjools: are you sure that's right? [04:20] massserver_node or maasserver_node [04:20] it through an error [04:20] oops [04:20] DiabolicalGamer: in the dbshell, update maasserver_node set owner=null; [04:21] :) [04:21] okay that worked lol [04:21] so should that fix it or do I need to change anything else? [04:21] bootstrap again [04:22] still no go [04:22] same error when bootstrapping [04:22] destroy again [04:23] and restart maas [04:24] * bigjools has to run out in a couple of mins [04:25] rebooting... [04:26] if that doesn't work, delete the nodes, re-commission, apply the sql above and then bootstrap [04:28] the command for the dbshell isn't working anymore [04:28] what does it say?> [04:28] and I have to go now unfortunately, I'll be back in an hour [04:28] ERROR: column "owner" of relation "maasserver_node" does not exist LINE 1: update maasserver_node set owner=null; [04:28] DiabolicalGamer: the bug as a better workaround [04:29] DiabolicalGamer: and an updated maas-provision is building in the testing ppa [04:31] cool, glad to hear it :-) [04:32] can you link me to the testing ppa? [04:39] is it possible to clone packages on another server? [04:39] build a cluster of servers with the same software [04:43] blendedbychris: yes but what you really want is config management [04:43] blendedbychris: not just packages.. config files.. everything [04:45] blendedbychris: for that, there's juju, puppet, chef, cfengine. :) juju is more than config management.. but its hard to explain :) [04:46] And does juju require an interpreter be installed on every node? :-/ [04:46] grumble puppet grumble [04:46] twb: juju uses an agent [04:47] currently written in python, though its being rewritten in go [04:47] Haha [04:47] bandwagons are go [04:47] go executables are standalone, so no interpreter then :) [04:48] twb: I haven't really seen anybody on the "go" bandwagon other than us juju interested people. :-P [04:48] Oh, wow, I didn't know Pike and Thompson were co-authors. They should advertise it as P&T Go, not Google Go [04:49] sako: aka plan 10. [04:49] bah [04:49] twb: ^ [04:49] sako: sorry. [04:50] SpamapS: i have been using csync for configuration management [04:50] think it's worth the effort for two servers? [04:51] (if you've already halfway configured the otehr [04:51] Though I won't be really happy until scheme (with opt-in mutable cons cells) takes over the world... [04:52] lifeless: neither of them are listed first on wikipedia though [05:01] DiabolicalGamer: you can upgrade to the testing ppa [05:28] anyone here use lsynd? [05:28] I have no idea how to configure the keypair [05:49] hi all [06:11] is there a csync2 alternative… i fricking hate setting up inet.d an djunk [06:11] and* [06:11] there should be a tool as easy as lsyncd [06:13] blendedbychris: I didn't understand a word of that. [06:15] csync2 configuration involves installing something for xinet.d and configuring private/public keys beyond what's already root. [06:15] just looking for an alternative [06:16] puppet seems to be a popular one [06:19] it doesn't run as a daemon i think is why i find it so difficult to install [06:20] blendedbychris: you're looking for configuration synchronization? [06:20] yes [06:20] just something that will copy config and restart the appropriate service [06:22] is there a way of removing 2 drives from a pool without replacing them? I have 2 IDE and 4 SATA drives. The IDEs are throwing errors, and I suspect slowing down the pool [06:22] bcfg2, puppet, and cfengine are the traditional approaches to that, but all may be a bit more complex than you're looking for [06:23] puppet has a daemon (puppetmaster) that serves configuration management to its slaves (puppet), which are IIRC invoked by cron [06:24] twb: Well, the intended approach for Puppet is to also run the client/slave as a daemon, and then have it configured to do its thing with regular time intervals, but it is also possible to run the puppet client standalone from cron. [06:24] One popular way of running puppet is entirely by hand -- you write the "manifest" with an editor, copy it to the host in question, and run puppet directly on it [06:25] Yeah, that second approach fits nicely with having the manifests in a git repo, on a nfs share, or so. [06:28] haha, my recorded bitching about puppet, was with you [06:28] http://paste.debian.net/163630/ [06:29] blendedbychris: how many boxes do you have? [06:30] 2 haha [06:30] blendedbychris: it is definitely not worth the effort [06:30] blendedbychris: do it by hand [06:30] (FSVO it = configuration management) [06:30] i've done csync2 before with two nodes [06:30] it was helpful [06:30] IMO you need >16 hosts before it's worth caring [06:31] i don't get why they intend it to be so damn "secure" [06:31] blendedbychris: because it amounts to a privileged RPC [06:32] my thoughts are … shared root keys with lsyncd is just as insecure [06:32] or rather privs [06:33] I am not familiar with lsyncd, but obviously in assymetric cryptography, it is a failure /a priori/ to share/reuse private keys between principals. [06:33] no idea what that means [06:34] all i had to do is ssh-keygen and ssh-copy-id on root to get it going [06:35] blendedbychris: you don't know what you're doing, and if you try to deploy ANY kind of configuration management, you are going to hurt yourself and probably others. [06:35] indeed [06:38] twb: Well, at least we ended up agreeing on something :) [06:39] glad we are all learning here [06:44] is there a way to remove a disk from a ZFS pool? [06:47] jeeves_moss: NFI, have you asked #opensolaris or #solaris? [06:47] Hell there might even be a #zfs [06:47] twb, not yet, but I bet no one is there tonight [06:48] twb, thanks. I've asked in all 3 channels [07:00] hi all, I want to setup VPN with the existing ipstar dish in each remote office sites like a star topology. Can accomplish using ubuntu server? === smb` is now known as smb [07:13] two USB hard drives on a home Ubuntu server, 1TB and 2TB; I need to have the full 3TB... plan on using them as just a means to do an identical mirror of just ONE USB3 drive connected to a different computer in the home, over the network. Trying to decide on the level of RAID I setup using mdadm (linux raid) ... I'm going between RAID0 (obviously no redundancy, but better IO and performance [07:13] maybe) vs. non-raid JBOD (straight concatenation, again no redundancy, but if one fails I don't lose all) ......... thoughts? [07:16] So wait, you actually have one each of 1TB, 2TB and 3TB, and you want to RAID0 the first two, and then mirror that with the 3TB? [07:16] sort of yeah, lol [07:17] nvm, just realize raid0 would not give full 3tb, dumb question: so if I do go about my actions it would be jbod I need to go with [07:17]