[00:02] RoyK: I've lived with and loved it since 1985, but every command line program writer comes up with their own sets of options and there are a number of different ways to specify them (single dash, double dash, single letter, word, space between option and arg, "=" between option and arg, some programs even use the DOS "/" to prefix options or no prefix at all, ...) [00:02] well, true [00:02] , does "-v" mean "verbose" or "version", does it process stdin by default or do I need to specify "-" or "-stdin", etc [00:02] haven't seen the DOS prefix, though [00:02] gnu seems to standardise most of their stuff pretty well, though [00:03] I'm not even saying it's bad. The flexibility and chaos have helped let great things thrive. [00:03] I've "heard" (ahem) that the VAX VMS command line was very consistent, but not much fun to use. [00:04] erichammond: that might be because it was written by a company instead of by hundreds of grad students at a bunch of different universities [00:04] In the "testdisk" package on Ubuntu, there is an effective utility named "photorec" which uses /d as an option specifier. [00:04] erichammond: heh - we still have two VMS boxes at the office :) [00:04] alpha stuff [00:11] And dd and wodim still use IBM-style [00:16] New bug: #506771 in clamav (main) "I can't download anything " [Undecided,New] https://launchpad.net/bugs/506771 [01:41] smoser: Thanks for the symlink on http://uec-images.ubuntu.com/releases/karmic/ [01:41] smoser: Could we also get one on http://uec-images.ubuntu.com/releases/hardy/ [01:41] yeah. i'll do that. i might get hassled about it though. [01:41] i did ask slangasek if it was ok to keep the old releases [01:42] and he said what i was doing was reasonable. i think this is reasonable too, but possibly others dont. [01:42] it should be possible for us to make the bundle downloadable from s3, right? [01:44] smoser: what bundle? [01:44] smoser: but yes, it's possible to make almost anything downloadable from S3 :) [01:44] and anything on S3 is automatically available through torrent. [01:44] hardy link should be there. [01:44] the published amis [01:45] ie, so people can get the from s3 rather than uec-images.u.c [01:45] Sure, then it would be faster and free as long as they were inside EC2. [01:46] its free right now for the consumer when it comes from uec-images.u.c [01:46] until june or something [01:46] No, the EC2 instance is paying for incoming network traffic. [01:46] oh [01:46] yah [01:46] and i could definitely see that being almost indefinite. makes sense for "move in special" [01:46] It's not that big of an issue for me, anyway. [01:47] yeah, its pennies anyway you look at it. but faster would be nicer. [01:47] and not taking canonical bandwidth too [01:47] anyway [01:47] hardy link should be there now [01:47] When I ran the script to generate the 12 EBS boot AMIs, it only took about 4 minutes each from start to finish. [01:48] Of course there were instances in each region, so I should have parallelized it. [01:48] next time... [01:49] lftp rocks. just in case you werent aware. i saw large increases using 'pget' [02:18] how do I change to a static ip in this thing [02:18] on what thing? [02:19] maxfiles: edit /etc/network/interfaces, change dhcp to static, and fill in the other fields. [02:19] maxfiles: 'man interfaces' for syntax details. [02:19] I switched to 8.0.4 lts server with desktop installed === erichammond1 is now known as erichammond [02:32] interesting issue I had on reboot :( [02:33] if I specify dhcp for the network interface, it ignores the mtu setting [04:35] I put MAILTO="" in my crontab file; but I'm still getting mail from the output.... [04:35] any ideas why? [04:37] MTecknology: It's in the manpage: "If MAILTO is defined (and non-empty), mail is sent to the user so named." [04:38] erichammond: then how do I have it discarded? [04:43] MTecknology: redirect the output of whatever it is your running from cron to/dev/null. [04:44] It may be useful to just send stdout to /dev/null and have stderr still get sent to you, so that you're aware of errors in your cronjob. [04:45] sbeattie: I was doing that and it got ugly. I was told that should work.. so crap doesn't get sent through it [04:45] It's mostly all ssh login messages; I had a few megs of just that [04:45] err, what cronjobis sending you that? [04:45] err, what cronjob is sending you that? [04:47] is it something that sends mail directly, a la logwatch or logcheck? [04:48] some backup scripts I have [04:48] some sync scripts [04:50] I guess I can tack on ' > /dev/null' to about 50 jobs; I just didn't want to [04:51] hrm, if it's actual cron output, that seems to be a bug, or lying documentation. [04:55] MTecknology: your problem is too many ssh attempts in your logs? [04:56] twb: no.. [04:57] twb: the cron events do a lot of ssh work, the text that you see when you log into the server winds up in /var/mail/.. [04:58] MTecknology: consider control master, then [04:58] twb: hm? [04:58] (It's a feature of OpenSSH to run multiple commands over a single shared SSH connection) [04:58] Or maybe investigate something like puppet, depending on what you're doing. [04:59] I'd still get the junk [05:04] MTecknology: you'd get one connection per instead of one per ssh command [05:04] is anacron or some other alternate cron involved? [05:04] Where might be the extent of the cron job, or the extent of the uptime [05:04] MTecknology: Where exactly is the "junk" coming from? A Banner statement in sshd_config? /etc/issue.net ? Other? Can you tackle this by preventing the generation of that banner info when the logon is not interactive? [05:05] jmarsden: that's part of the junk, put there's other output from other tasks like that which I don't want to get [05:06] MTecknology: Well, if you'd reduce the volume by 50% by killing the login banners, that would be a start :) [05:07] I have MAILTO="admin@domain.com" now too... at least it'll show up where I'll actually read it [05:07] MTecknology: it would be better to alias root to admin in /etc/aliases [05:09] twb: so the top of that I do root: mail@domain.com ? [05:10] or, the end [05:10] Sure [05:10] Then run newaliases [05:12] twb: interesting - thanks [05:13] jmarsden: thanks for that idea; how can I do that? [05:14] MTecknology: Well, that depends where the banners are coming from, hence my earlier question... I'm not 100% sure how but once we know what generates the text we can look at ways to stop it being generated for your cron jobs... [05:15] jmarsden: /etc/issue.net [05:17] OK. Is there a statement Banner /etc/issue.net in sshd_config, or is something else causing that file to be output? [05:19] If it is a Banner statement I'm wondering about making that conditional on a Match in there... not sure if it will work, but worth a shot... [05:20] jmarsden: that's the line [05:22] OK, so now you need to find what is different about the ssh connections being made by your cron jobs compared to those being made by interactive users, and then create a Match block with the Banner inside it for the interactive users. [05:22] man sshd_config for more on Match blocks... [05:22] alrighty, thanks [05:23] You're welcome. [06:06] If i have a RAID 1 array on one computer, and i take out a drive and stick it into another computer (non-RAID), will the second computer be able to read it correctly? [06:06] cyphase: RAID-1 is a mirror. [06:06] twb: i know, i'm just making sure [06:06] cyphase: at least traditionally, that meant that both drives are directly mountable [06:07] You'll want to mount -oro to avoid getting the array out of sync, though [06:07] twb: right, that's what i thought.. thanks :) [06:07] This is assuming you're talking about md RAID [06:07] twb: i don't know what kind of RAID it is yet [06:07] doubtful it's linux though [06:08] I can't comment on fakeraid or hardware raid [06:10] twb: okay, thanks for the info [06:19] MTecknology: You could also send the email to "devnull@yourdomain.com" and then add an alias for "devnull: /dev/null" [06:20] though I didn't quite follow why you couldn't redirect everything to /dev/null in the cron job itself. [07:10] hello, i need help, i install linux server 9.04 x64bit and i want to change block size 4096 to 512 . how can i do this ? [07:10] you need partman I suppose [07:11] so all data lose ? [07:11] if i change block size to 512 ? [07:11] don't really know [07:12] you'll have to reformat to change the block size === FireCrotch_ is now known as FireCrotch [07:58] Is it possible to use dpkg-divert for a whole directory? I want to divert /usr/share/wine/fonts/ to /usr/share/wine/english-only-fonts/... [08:26] good mornign [08:26] I have a server with one NIC and a vpn interface. This server hosts about 14 sites, out of which i want 10 to be available only on vpn network and 4 to public. I need a combination of IP and Name based vhosts in apache, any ideas? [08:30] Hi, running Ubuntu Server 9.10, kacpid is using a LOT of cpu. How can I stop it safely? I see no /etc/init.d/acpid [08:33] <_ruben> error404notfound: NameVirtualHost directive takes an ip address as parameter which can be used to make the distinction between vpn and public .. then servername/serveralias for the actual vhosting [08:33] <_ruben> psteyn: its a kernel process, the only way to stop it (probably) is to disable acpi at boot time [08:33] _ruben, yup. got it, thanks to SpiceMan on #httpd [08:35] psteyn: You can disable acpi by appending acpi=off to your GRUB kernel line and rebooting. Be warned however that you'll lose features such as CPU throttling, fan control, etc [08:36] Meh. :( I don't really mind losing that, but I've also read that acpi=off will disable SMP [08:37] I do mind that.. [09:19] how to uninstall a program installed with "make ; make install" ? [09:20] <_ruben> depends on the program [09:20] <_ruben> some offer the option of running make uninstall [09:20] <_ruben> otherwise you might have to delete all installed files by hand [09:40] _ruben, but how do you know the list of files installed ? [09:41] _ruben, and what if an existing file has been changed by make install ? [09:43] <_ruben> analyze the Makefile is one way .. there's also (wrapper)scripts that do that for you (and possibly even offer uninstall functionality) [09:43] martin-: dpkg -l [09:43] sorry [09:43] <_ruben> if files are overwritten (without being backed up), you wont get them back (easily) [09:43] maxagaz: dpkg -l [09:44] oh, and sorry again, I just read back [09:45] now I put a lock on my mouth [10:16] i'm trying to keep a static IP on my eth1,which is also the interface from which the dhcp3-server gives out IPs,but the static IP keeps getting overwritten by dhcp,thereby messing up my lan: http://dpaste.com/151096/ [10:16] do I need dbus on an 8.04 apache server? [10:18] dayo: dhclient.conf supersede options [10:19] and what with dhcp server? dhcpd eth1 and dhclient eth1 ? oO [10:22] i have port 113 closed on my ubuntu server, what if i want to put it under stealth/reject all connection on this? would it effect any running services? [10:25] mealstrom: what do u mean? [10:26] New bug: #513151 in mysql-dfsg-5.0 (universe) "package mysql-server-5.0 (not installed) failed to install/upgrade: subprocess new pre-installation script returned error exit status 1" [Undecided,Incomplete] https://launchpad.net/bugs/513151 [10:27] <_ruben> error404notfound: not making much sense (atleast not to me) [10:28] _ruben, port 113 on my server appears to be closed if i do portscan. Okay? what if i explicitly configure iptables to "reject" any connections on this, that way it won't even appear in portscan. Will rejecting any connection on this effect any running services? [10:30] <_ruben> closed and reject are or more or less the same thing (one results in the other) [10:30] hi [10:30] <_ruben> drop -> filtered ... reject -> closed .. accept -> open [10:31] _ruben,hmmm [10:31] I've grown the volume of a virtual disk in a virtual machine, and I need to grow also the LVM partition. How can I do that? lvextend? (it's the physical disk, I guess) [10:32] <_ruben> acalvo: increase pv, vg, lv and then fs [10:33] _ruben: ok, I'll try it [10:33] and thanks, you're always here to help :) [10:33] <_ruben> just dont tell my boss, he thinks im working ;) [10:34] hmmm, okay, have filtered that as well, now i am wondering what next should i do with this server. :P [10:35] quite confused: I've this [10:35] Disposit. Inicio Comienzo Fin Bloques Id Sistema [10:35] /dev/sda1 * 1 621 4988151 8e Linux LVM [10:35] /dev/sda2 622 652 249007+ 5 Extendida [10:35] /dev/sda5 622 652 248976 83 Linux [10:35] I've increased the vmdk 1G [10:35] how do I apply that? [10:35] I mean... the pvresize applies only to a partition [10:36] while is the whole disk whichs is bigger [10:37] I just want to make /dev/sda1 1G bigger [10:38] acalvo, !pastebin [10:38] error404notfound: mmm ok [10:44] <_ruben> acalvo: you cant make sda1 bigger (easily) [10:44] <_ruben> however, you could use the benefits of lvm [10:45] <_ruben> the extra 1G can become a new partition, which will be a pv, which you can add to a vg, which allows you to create extra lvs or extend one or more lvs [10:51] _ruben: I know, and it's quite useful [10:52] however, I'd like to extend the sda1 [11:01] acalvo: Why? [11:02] acalvo: You can create a new pv at the end, add it to the same vg as sda1 and the you will be able to allocate the space in there with lvcreate just like you normally would. [11:06] soren: ok, that makes sense, however, I'm not that into LVM to understand everything. If I get it right, I just have to: first, add new size to the vmdisk of my virtual machine; create a new pv, add it to the same vg as the other lv, and then it'll be as if the /dev/sda1 (in my example) were bigger? [11:06] Umm.. [11:07] Well, apart from the bit about /dev/sda1 being bigger.. [11:07] You vg will be bigger. [11:07] aha, ok, so I'll be able to create more partition (in a traditional way) [11:07] Right. [11:08] but not extend the current partitions (again, traditional way) [11:08] Right. [11:08] You add a new partition that takes up the newly allocated space. [11:08] pvcreate it [11:08] vgextend your vg with the new pv [11:08] Which virtual server implementation would be good for Sparc? [11:08] I've tried vserver etc [11:08] ..and that's it. Your vg will be bigger, and you will not need to worry about it being split across multiple partitions. [11:09] acalvo: It's probably easier, though... [11:09] acalvo: ..if you just add another virtual disk. [11:09] soren: don't worry, you've clarified lots of things [11:09] acalvo: Instead of bothering with extending the existing one. [11:09] right! [11:09] acalvo: The effect will be the exact same. That's part of why LVM is so awesome :) [11:10] I understand [11:10] Great. [11:10] thanks soren! I'll try that! [11:10] Just shout if you stumble. [11:10] * soren heads to lunch [11:27] soren: I've managed to do it by extending the vmdisk (however having a new virtual disk seems fine) [11:28] so now my /dev/vg/lvname has more space [11:28] I didn't get it the first time [11:55] Kerberos with OpenLDAP backend config ? - I know some about both kerberos and LDAP, but how to tell kerberos to user LDAP as backend. ?? [12:21] * soren curses the launchpad UI for copying packages between PPA's [13:06] I am trying to run fsck on a remote machine. The disk I am trying to check is not the boot disk. It tells me it is busy, but is not mounted: http://pastebin.com/m43085014 What can I do? [13:09] Is there a test suite for likewise? I'm working on porting it to ARM, and would love to make sure I can test everything off it [13:09] Italian_Plumber: Can you pastebin the output of mount, please? [13:14] http://pastebin.com/m7a468855 [13:15] Italian_Plumber: and what version of e2fsprogs are you running? [13:15] (dpkg-query -W e2fsprogs) [13:15] Italian_Plumber: /dev/sdb1 is mounted [13:16] but when I try to unmount it, it says it's already unmounted. [13:16] Italian_Plumber: No. [13:16] Italian_Plumber: When you try to unmount /dev/sdb, it does. [13:16] Not /dev/sdb1. [13:17] ah. [13:17] david@warthog:~$ sudo umount /dev/sdb1 [13:17] umount: /media/data: device is busy [13:18] hey guys. I'm using the 9.10 EC2 AMI (32bit = ami-bb709dd2). I've noticed that sources.list (/etc/apt) does not include multiverse by default. is there a simple way to add multiverse to apt without having to edit the sources.list file manually? [13:18] Italian_Plumber: sudo lsof /media/data [13:19] also make sure you're not in the mount point [13:19] when unmounting [13:24] * Determinist blinks [13:24] is there no simple way to add multiverse to the repos aside from editing sources.list directly? [13:27] OKay, I apparently have successfully unmounted it now, but now I get this: http://pastebin.com/m15218c40 [13:29] Italian_Plumber: pastebin output to 'sudo fdisk -l' i guess [13:29] Italian_Plumber: or just 'sudo fdisk -l /dev/sdb' [13:31] here is the output to several of the commands: http://pastebin.com/m33bc1a41 [13:32] Italian_Plumber: why are you tring to check this filesystem anyway? [13:32] *trying [13:32] Determinist: You're making it sound like that's super complicated? [13:32] I get the "superblock" erro when starting the machine, which means the machine won't start up without interaction. [13:32] Determinist: You open the file, add multiverse at the end of the line, done. [13:33] Italian_Plumber: could be that the superblock is corrupted. you may need to specify a backup one (with the fsck command) [13:34] Italian_Plumber: you see an example of doing so in the output to your fsck command [13:34] Italian_Plumber: but you should research where the backup superblocks are found in my opinion [13:36] with the fsck command, Iget this: http://pastebin.com/m5c15f459 [13:36] New bug: #512646 in postfix (main) "Could not install postfix - postinst failed to properly create postfix user" [Medium,Incomplete] https://launchpad.net/bugs/512646 [13:36] sorry I thought I included that in one of my pastes, but I guess I didn't. [13:39] Italian_Plumber: again, you're not using the proper device name [13:40] I get the same output with "sdb1" [13:42] Italian_Plumber: pastebin that ouput along with mount output again [13:45] sure... thanks.. http://pastebin.com/m3ffd48ff [13:49] soren: yeah, that's fine when you have 1 machine. but when you have a load of them on EC2, things get more complicated. i'm looking for a way to automate this. if there's no simple way, i'll just write a script to do it. [13:52] smoser: thanks for the uec-tools for bundling and uploading a complete image tarball, will rock my UEC training today :) [13:54] good. i hope to have it packaged in lucid. [14:03] Italian_Plumber: maybe specify the block size with e2fsck [14:13] kirkland: are you able to complete a fully automated install with today installer? [14:13] mathiaz: will get back with you once i try [14:14] kirkland: context -bug 512632 [14:14] Launchpad bug 512632 in debian-installer "Network component not activated on a fully automated installation" [Undecided,New] https://launchpad.net/bugs/512632 [14:14] mathiaz: preesid [14:15] this bug is easily fix with apt mirror 512632 [14:15] hey!! [14:16] I've setup a webserver with ubuntu 8.04 server edition [14:16] deb-amd64 http://archive.ubuntu.com/ubuntu karmic main/debian-installer restricted/debian-installer universe/debian-installer multiverse/debian-installer [14:16] but after I leave it idle for a long time it becomes inaccessible by ssh [14:17] apache is also not accessible...though i am able to ping it [14:17] does anyone have a clue what might be the issue [14:17] mealstrom: what do you mean? [14:17] anyone? [14:18] hello, would anyone know why my ubuntu server often decides to suddenly switch from static to dynamic IP? [14:19] mathiaz: when you are using local mirror (apt-mirror) for network installation -- there files from debian-installer aren't mirrored and you've got errors with *.udeb files [14:19] mealstrom: well - I'm not using a local mirror [14:21] try to add netcfg/get_hostname=ubuntu [14:25] How do I make sudo update the environment? It seems like sudo -u www-data command from the crontab doesn't update that. [14:28] MTecknology: i believe the sudoers file contains stuff for that (man sudoers) [14:29] pmatulis: I looked through that already [14:30] it runs the command as the user; it just doesn't update environment [14:43] Is there a method to know when an lvm snapshot was created? (I want to search for the oldest snapshots) [14:58] jjohansen, where do i get the ec2 kernel source? there are no tags in git://kernel.ubuntu.com/ubuntu/ubuntu-karmic.git for 2.6.31.302.2 (which is the current version of linux-ec2 package in karmic) [14:59] smoser: the import last time dropped the reference. I try to put it in the commit log and under debian.ec2/patches.xen [15:00] the update I am working on pulled in fresh yesterday [15:00] where do i get debian.ec2/patches.xen ? [15:00] oh. on that branch [15:00] yeah ec2 branch [15:04] jjohansen, where do i get the ec2 branch? [15:04] smoser grab the git tree [15:04] got it [15:04] at least i have the one i pointed at above [15:05] then do a git checkout --track origin/ec2 -b ec2 [15:07] ah. its been a while, too much bzr [15:07] forgot about remote branches [15:07] soren: have you used the new kvm in lucid much yet? [15:07] soren: the 0.12 one? [15:07] kirkland: Some. [15:07] kirkland: Not much. [15:07] soren: any major breakage? [15:08] jjohansen, now, where do i see/get -302.2 ? [15:08] kirkland: Not that I've seen, no :) [15:08] soren: minor breakage, then? :-) [15:08] smoser: git log, find the commit that mentions bump to -302.2 [15:08] the git checkout the hash for the commit [15:08] kirkland: Nothing that I really can attribute to kvm, no. :) [15:09] there is no tag [15:09] soren: nice [15:09] the only 302.2 htat i see is in checksums [15:10] smoser: give me a sec [15:15] smoser: where did you get a 302.2 from? [15:16] $ dpkg-query --show linux-ec2 [15:16] linux-ec2 2.6.31.302.2 [15:17] right, but is this a local vm, an ami on ec2? [15:18] I am wondering if it is one of my test kernels [15:18] offiical released [15:19] * jjohansen is baffled [15:19] so we have lost some history [15:20] that shouldn't happen but it is possible to do when rebasing [15:20] the ec2 kernel, gets rebased against the master kernel branch [15:21] we are supposed to keep all commit info but I could see that getting dropped [15:22] I can tell you that the version of xen patches hasn't changed since earlier than that [15:23] smoser: what do you want to know about the patches? [15:23] smoser: it looks like we are using the patches from 20090902 [15:24] mostly i was just interested in config changes. i'm tryiing to buidl a ext4 kernel module. [15:24] and also just wanting to walk through this once [15:24] smoser: you appear to not be on the team call... [15:24] ah [15:24] 2 penalty laps around the court yard! [15:29] I still can't figure out how to ignore the SSH banner for non-interactive logins.. [15:30] jjohansen, easiest way to build kernel modules ? without building entire thing [15:31] assuming the directory has a Makefile that can do it, [15:31] cd into the directory [15:31] make -C -M=`pwd` [15:45] zul, mathiaz: who takes care of mysql among you? [15:46] mdeslaur: mathiaz and me [15:46] mdeslaur: whats wrong? [15:46] zul, mathiaz: could you two come to #mysql-ndb [15:46] apparently the clustering in 5.1 that we ship is old and deprecated [15:47] yeah we are shipping mysql-cluster 7 in lucid [15:51] New bug: #511180 in samba (main) "File sharing failed" [Undecided,Invalid] https://launchpad.net/bugs/511180 [15:51] mathiaz: can you join #mysql-ndb please? [16:06] here's first papercut: kill php's error display by default, put them in apache's error logs [16:06] :) [16:10] hmm, i wonder what search engine i should use [16:15] ivoks: Your pacemaker/apache/vsftpd test cases work perfectly on my test machines. 3 node/ESX. [16:15] Omahn: that's awesome [16:15] Omahn: could you add comments to wiki page? [16:16] https://wiki.ubuntu.com/ClusterStack/LucidTesting#Test results [16:16] Indeed. I'm just testing the drbd test case. The resync is taking a while :-) [16:16] hehe [16:18] there might be some issues with that test [16:18] drbd upstream said they'll send me modifications for it [16:19] hi [16:19] when setting up a master-slave relation with bind [16:19] does the slave has to have its own zone files with the @ set up to it sip address? [16:19] or it has to have the same as the primary? [16:20] it should pull everything from master [16:20] *everything*? [16:20] domain records [16:21] even this? @ IN SOA mars.esci.es. root.esci.es. ( [16:21] of course [16:21] ok [16:21] and this too? @ IN NS mars.esci.es. [16:21] yes, imo [16:21] I've thought it was poiting to the machine runnig the dns server [16:21] wait, i'll check [16:22] ok! [16:24] yep [16:24] it creates zone file [16:24] ok, it points to the primary master [16:24] (authority) [16:29] ivoks: DRBD test case works fine for me too. Was issue was you expecting? [16:29] *What [16:29] i had no issues [16:30] mathiaz: " kirkland: not enough there to be sure, but there's an outstanding console-setup bug that multiple people are running into and that ev's looking at, so if it's lucid then that's a possibility" [16:30] but upstream said they have some comments [16:30] i don't know what exactly [16:30] Ah I see. Anyway, works fine for me :-) [16:30] I'll update the wiki page. [16:30] great [16:30] kirkland: are the preseed install also failing for you? [16:31] kirkland: I've asked cr3 to do a test install and it works well for him [16:31] mathiaz: they're hanging at a screen asking me to select the next step in the process [16:31] mathiaz: if i select "configure the keyboard", nothing happens, i just loop on that step [16:31] kirkland: right [16:31] mathiaz: if i push it to the next step, detect nw hardware, it continues [16:31] kirkland: and then if you select configure network, it works [16:31] kirkland: awesome - that's what I see as well [16:31] mathiaz: sounds like cjwatson and ev are aware of the issue [16:32] kirkland: great - and that wasn't case before you left for LCA? [16:32] mathiaz: bug #512592 [16:32] Launchpad bug 512592 in debian-installer "console-setup-udeb succeeded but requested to be left unconfigured" [Undecided,New] https://launchpad.net/bugs/512592 [16:32] mathiaz: correct [16:32] mathiaz: this worked liked a champ before I left [16:35] Omahn: you had only two servers in drbd test? [16:35] ivoks: Only 2 taking part, I had a 3rd running with corosync up but not involved in drbd. [16:35] right [16:35] that's the test i wanted to see :) [16:36] I think I tried initially with 2 only, and then realised my earlier quorum setting required at least 2 nodes to be up. [16:36] so, we can say that we have good replacement for rhcs [16:36] So I did it again with 3. [16:36] yes, it won't work with 2 [16:36] 2 node clusters are pointless [16:37] they do more harm than good [16:37] I've never tried RHCS but Pacemaker/corosync seems to work really well for us so far in testing. Just a bit tricky to integrate with Puppet but i'm getting there :-) [16:37] Actually our new LVS clusters will be 2 node. [16:37] and what happens when they lose interconnection? [16:37] :) [16:38] they both thing other one is dead [16:38] think [16:38] and then you have big mess [16:38] 3 is minimum [16:39] 2 + stonith should be ok presumably? (And as these are routing packets, if they lose connectivity then we have much bigger problems) [16:39] 2 + 1 something to keep track which one is really dead [16:40] it could be nokia n900 :D [16:40] Ok I'll have a look at running 3 nodes. [16:42] so, we now just have to write MIRs for that stack [16:43] jiboumans: do we have a green light for corosync/pacemaker cluster in main? :) [16:43] Fingers crossed :-) [16:44] Omahn: http://clusterlabs.org/wiki/DRBD_HowTo_1.0 [16:46] ivoks: Thanks [16:47] i'll merge latest drbd and pacemaker parts now [16:47] ivoks: Good stuff. I'm building an LVS cluster tomorrow on real tin so that would be fantastic. [16:53] When an interface is brought up is any device created to check for its existance in a script or is the only way by parsing ifconfig output [16:53] My grammar was shocking there [16:53] Is any device created that allows me to check for its existance with a script [16:55]