[00:01] thanks [00:07] Xinu! [00:07] KillMeNow: hey boss, what it is? [00:08] trying to not take a nap while at work [00:08] at least not for a little bit [00:08] LOL [00:08] KillMeNow: that's the way to be === monteith_afk is now known as monteith [01:31] i have ubuntu on 3 different machines with openssl server. how come on some the login responds instantly and some it pauses for about 15 seconds before asking for the password? [01:31] system load. [01:32] in my case, my router was being used for an SSH man-in-the-middle attack [01:33] nah, its not system load [01:34] im not saying it cant be, but that it is not the case for me. [01:35] and it is consistent, one machine is immidate, the other two are not. all upgraded to the latest packages. === Debolaz is now known as Guest80595 [02:06] smoser, soren: I happened to scan the logs and saw some discussion of bug 431255. [02:06] Launchpad bug 431255 in ec2-init "ec2-init: Move ec2-run-user-data to startup priority S99" [Undecided,Confirmed] https://launchpad.net/bugs/431255 [02:06] soren is long since asleep [02:06] but i'm here for a bit. [02:07] i was really hoping that soren would summarize in the bug report [02:07] I agree that there isn't much difference between S71 and S99. I don't think I had much of any particular reason for S71. [02:07] I probably copied somebody else's S70 for updating authorized_keys and then just went one higher. [02:08] I'm hoping that the last few comments in the discussion did not imply that this bug would not get fixed for Karmic. What was your understanding there? [02:08] i think the net was that we don't see a lot of reason for making this change now. wanting to be able to log in and watch user data run isn't a huge deal in and of itself. [02:08] i really dont want to regress [02:08] i think thats where soren was standing [02:08] i really just see a lot of things to get done, and dont want to regress anything [02:09] The problem is that it breaks code which is already running on images on EC2. I would have to rewrite some of my code to get it to work on Canonical's images. [02:10] Nobody is running production code on Canonical's karmic images yet, so they should start out with the de facto standard, especially since everybody seems to agree that's the ultimate direction anyway. [02:10] CppIsWeird: Check your dns settings on the slow ones [02:10] what would break? [02:10] see if they can resolve your hostname, and their own [02:12] smoser: Users drive the setup of an instance in a number of different ways. One of my production use cases has the user-data script wait until an EBS volume is mounted, but the process which attaches the volume first has to do some stuff on the instance through ssh. This deadlocks and both processes are waiting for each other. [02:13] (since ssh is not available until user-data completes) [02:13] can you write that in the bug ? [02:13] Sure thing. [02:14] A simpler example might be: User does not want to put private keys in user-data, so user-data script waits until they show up on the file system. External process scp's the keys in after starting the instance. [02:15] do do do === Guest80595 is now known as Debolaz [02:35] Interesting: http://www.channelregister.co.uk/2009/09/21/yahoo_shopping_zimbra/ [02:36] zul: done done done (?) [02:36] i wish [02:38] smoser: Added comment to bug 431255. [02:38] Launchpad bug 431255 in ec2-init "ec2-init: Move ec2-run-user-data to startup priority S99" [Undecided,Confirmed] https://launchpad.net/bugs/431255 [02:44] smoser, soren: I would be happy to do the work to split the ec2-init sh script using an approach which had the least chance of breaking anything, if my branch had a chance to be merged. There would be duplication of 16 lines of code (run_once/run_once_per_ami) but that's about the only downside. [03:09] Bah, damn server keeps locking up. [03:20] What is the correct way to recover a kernel panic message after a reboot? (Or enable recovery of it the next time it happens) [03:29] Debolaz: I [03:31] Debolaz: It's been a long time since i needed to work at that level. I used a serial port to collect the msg and a core to debug. [03:33] This is a remote server so I'm hoping to be able to get away with something more convinient than a serial port for the first debugging attempt. That being said, if it has to be done, it has to be done, but could the kernel be configured to relay this information over TCP or UDP instead? [03:36] yes ... let me look [03:37] Debolaz: lookup syslogd. [03:38] Debolaz: or logserver http://www.aboutdebian.com/syslog.htm [03:39] Will 10.04 switch to rsyslog, as Debian did in Lenny? [03:39] Plain syslog blows [03:39] twb: We have switched for 9.10. [03:40] ScottK: cool. [03:40] (I only track LTS, you see.0 [03:40] nick_schembri: How do I configure the kernel to log directly to this though? If the kernel is panicing, it's not going to return to userland, and hence the message won't be forwarded to the remote server by syslogd. The kernel has to do it directly itself. [03:41] Currently in 8.04 I send syslog UDP packets over an SSH -w TCP VPN, because it seemed like a better idea than installing a non-main category syslog :-( [03:41] The next iteration of my products will use a proper OpenVPN UDP VPN. [03:46] Debolaz: not sure. if the system is faulting before it can write to a file or send one udp packet, you are going to need to use a serial port to get as much as you can before it stops. [03:48] nick_schembri: But it would still be able to send a network message out. The kernel has access to the NIC and network stack, just as it has access to the serial port. There's nothing technically preventing it from sending out a message. However, a panic does mean that userland is no longer a safe area, and it will never return to that place, making it technically impossible for syslog to send a message to a remote server. [03:55] Debolaz: will the system run for a time? [03:55] Debolaz: can you ping the system after it faults. [03:58] No, there's no response at all from it. === erichammond1 is now known as erichammond [05:49] Hi, im trying to install smokeping on my ubuntu server(8.04).. when i run apt-get install command, i get a PAM authentication error [05:49] here's the error:Your account has expired; please contact your system administrator chfn: PAM authentication failed adduser: `/usr/bin/chfn -f SmokePing daemon smokeping' returned error code 1. Exiting. dpkg: error processing smokeping (--configure): [06:35] New bug: #433863 in image-store-proxy (main) "image-store-proxy reports failure to start while it starts correctly" [Low,In progress] https://launchpad.net/bugs/433863 [08:16] Hi guys. I think I'm hitting an ext4 bug which has been fixed in 2.6.29. However, my latest updates available take me to only 2.6.28 [08:17] What is the proper way to use a later kernel in ubuntu server? [08:17] Using 9.04 x86 [08:17] sorry 64bit. [08:18] <_ruben> kernels dont get backported, yet, i think i heard about plans to do so in the (near or distant) future [08:18] So..manual compile? [08:18] <_ruben> one option would be to use a mainline build (doesnt have the ubuntu specific stuff in 'em though), or backport a karmic kernel [08:18] <_ruben> manual compile would be worst [08:19] meh. [08:19] thanks :] [08:19] <_ruben> karmic is on 2.6.31 [08:20] can I get the deb src and create my own deb? [08:20] <_ruben> backporting isnt all that hard .. i think there's even a special tool for it .. prevu or something like that [08:20] <_ruben> sure [08:21] cool [08:30] I use newer kernels sometimes [08:30] check out the git repo, then run a packaging script, voila [08:31] (hardy has kernels that don't like my RAID controller) [09:09] having a small issue with the kernel that is default in Karmic Alpha6 it fails to recognise my IBM ServRAID 4LX has an issue that just says timeout ! so has older SCSI cards depreciated ??? [09:11] SCSI storage controller: Adaptec AIC-7892P U160/m (rev 02) RAID bus controller: IBM ServeRAID Controller [09:16] New bug: #434507 in nagios3 (main) "package nagios3-common 3.0.6-2ubuntu1 failed to install/upgrade: ?j pre-removal szkript alfolyamat 2 hibak?ddal kil?pett" [Undecided,New] https://launchpad.net/bugs/434507 [09:22] hi [09:23] how to restrict firewall with iptables?? [09:31] <_ruben> how to wait for answers?? [09:32] :D [09:32] to true! [09:44] Tejas was that a question? [09:45] It had a question mark at the end.. [09:45] ...but he left. [09:51] * soren takes a break [10:11] is there anyway to increase the number of inodes without reformatting? [10:17] Hi [10:17] in order to be able to send mail thru PHP in a ubuntu server, do I need to have sendmail installed if I'm have postfix installed in another server? [10:30] <_ruben> i'd install postfix locally (or another queue-based mta), so you wont lose any mail when the other server isnt available for whatever reason [10:33] mhh, install symfony using pear is not working well [10:34] _ruben: but it's always wise to send through an antispam-gateway, so never from the server itself [10:35] <_ruben> MatBoy: both actually .. local mta that has a spam-checking smarthost configured [10:36] _ruben: that is what I say [10:36] _ruben: it's actually a relay and not a smarthost anymore [10:37] <_ruben> MatBoy: not that much difference between the two in my eyes, then again, just blame microsoft for the term smarthost :) === ropetin_ is now known as ropetin [10:38] _ruben: hehe, you are never wrong ;) [10:38] _ruben: there is a very BIG difference actually [10:38] <_ruben> then again, i think postfix on ubuntu uses the term smarthost as well [10:39] mmm [10:39] maybe a silly question [10:39] _ruben: yes for n00bs :P [10:39] but, where can I change the smtp host variable for PHP? [10:39] _ruben: smarthosts are used in sendmail which is a bitch anyway [10:39] I think I have resolved the issue I was having earlier, it appears by using the boot option of "pci=noacpi" the SCSI storage controller: Adaptec AIC-7892P U160/m (rev 02) & RAID bus controller: IBM ServeRAID Controller are discovered and the system can boot! so looks like my hardware isn't truly ACPI compliment! [10:39] <_ruben> acalvo: most likely in php.ini [10:39] <_ruben> VK7HSE: lovely [10:40] acalvo: well, that's what I've thought, but it does not seem to get the new values [10:40] _ruben: not wrong! I was sweating it out a bit there for a while! ... ;-) [10:41] <_ruben> acalvo: talking to yourself eh :) [10:41] <_ruben> acalvo: did you restart apache? [10:41] well, most BOFHs do it [10:41] yes, reload and restart [10:41] but I'm quite confused with the only win32 comment [10:41] I know what it means [10:42] I just don't understand why they put it there [11:07] where is the ubuntu git repo? I want to get the latest kernel build from there [11:10] <_ruben> https://wiki.ubuntu.com/KernelTeam/KernelGitGuide [11:11] ttx: I'll roll a new Eucalyptus package before I go to lunch.. [11:11] psteyn: kernel.ubuntu.com [11:11] soren: ok [11:13] tnx [12:22] * soren lunches [12:23] again, I've a question [12:23] I'm still setting up a mailing list [12:23] in a mail server [12:23] everything seems pretty easy [12:23] but there is one thing I need to settle down [12:24] my server has various DNS names [12:24] in order to use mailling lists, should I change the hostname? is it enough to set up a DNS name? should postfix know any of this names? [12:25] it would be nice if postfix knows about the MX record [12:26] and this is setted up in the mydestination variable in the postfix main.cf? [12:27] #postfix [12:32] morning [12:49] helo all, I have a script that makes a movie. When i run it from command line as normal user, it works. But when i run it from cron as ordinary user, it makes empty file. i been folowing tips from here: https://help.ubuntu.com/community/CronHowto could some one help me please ? [13:19] szczym_: I'd suggest you ask in a support forum for the encoder you're using (like #mplayer if you're using mencoder, etc.). === psteyn_ is now known as psteyn === alexxxxx is now known as TANATHOS [13:49] cjwatson: I'm looking into eucalyptus failure to autoregister components right now -- Among other problems the latest eucalyptus refuses that you register things using "localhost", see bug 434593 [13:49] Launchpad bug 434593 in eucalyptus "euca_conf --register-walrus localhost fails in 1.6~bzr808" [High,Triaged] https://launchpad.net/bugs/434593 [13:50] cjwatson: I'm not exactly sure how we can workaround that though... which makes the whole idea of autoregistering at risk [13:58] soren: http://twitter.com/vrillusions/statuses/4151036378 [13:58] jpds: Nor cron. [13:58] Haha. [13:59] Well, it's Just Enough Operating System. That's the point. [14:00] ttx: maybe we could use CC_NAME for that, or maybe we could fix that restriction [14:00] <3 deadlocked LVM on database box [14:00] ttx: I'd recommend asking upstream why there's that restriction [14:01] cjwatson: Will do. soren suggests that whatever it is registered with is passed to other elements in the cloud [14:01] could be [14:02] It makes sense for both of them, really. They're infrastructure components needed by node controllers. [14:02] ...but it's still just guesswork. [14:02] I'm working on the (perhaps optimistic) assumption that it's a reasonable restriction. [14:03] cjwatson: RE: "use CC_NAME": if it resolves to 127.0.{0,1}.1 it will still be refused. It's not just a "localhost" pure match [14:04] don't do that then :-) [14:04] but CC_NAME doesn't really have to be a DNS name [14:05] so that probably isn't the right answer [14:23] * soren goes to pick up his daughter at day care === bitproph1t is now known as bitprophet [15:24] hi [15:25] some time ago someone suggest me a substituion for mailman, but I don't remember the name [15:25] <_ruben> ezmlm is the only other i know of [15:27] hi [15:27] F1 -> how can i see if proftpd is making querys to mysql correctly? [15:28] _ruben: I think it started with S [15:28] symphony [15:28] or something like that [15:28] but I'm not able to remember it [15:28] sympa! [15:28] ehehhee [15:33] smoser: ping [15:33] here [15:34] smoser: did you file the missing MIR for the UEC images ? [15:34] i don thave bugs yet, but spent the last day and this morning working on the 6 MIR. [15:35] i would love review of them, i will open bugs rsn [15:35] smoser: I can review that [15:35] smoser: just give me the pointers [15:36] k [15:37] https://wiki.ubuntu.com/MainInclusionEc2-Init https://wiki.ubuntu.com/MainInclusionEuca2ools https://wiki.ubuntu.com/MainInclusionPython-Boto https://wiki.ubuntu.com/MainInclusionCheetah https://wiki.ubuntu.com/MainInclusionConfigobj https://wiki.ubuntu.com/MainInclusionM2crypto [15:37] configobj is the one that i'm most concerned about. === BrixSat_ is now known as BrixSat [15:39] ttx, unless you object, i'm going to start filing bugs [15:39] smoser: no, please do [15:39] why this? http://pastebin.com/m7cc39b71 [15:40] i dont get it :s [15:42] smoser: MainInclusionCheetah > "upstreaam" typo. I'd just say as Rationale that it's a dep of ec2-init [15:44] smoser: MainInclusionConfigobj > needs some more meat, I guess. Ping zul if you need his MIRfiling expertise [15:44] hmmm? [15:44] i agree it does [15:44] its the hardest one to come up with [15:44] as there are other config options in python in main [15:44] zul, if you can offer time or help or whatever, please do [15:45] the MIR team is probably going to come back and say why dont you use the ones in main [15:45] smoser: the "Does upstream expect it?" is not about asking upstream developers about it. In your case the rationale is more about dependencies [15:46] what is "does upstream expect it" supposed to mean [15:46] you need a rationale for ec2-init... then for all the others the rationale is "dependency of ec2-init" [15:47] smoser: i don't really know -- sounds like a recent addition to the template (or one I always ignored) [16:02] sore, smoser,zul: meeting time [16:02] ack [16:03] o/ [16:36] kirkland: do you have a second to sound out an idea I have wrt byobu? [17:10] what is the regular guid for proftpd? [17:11] BrixSat, does it not run under it's own username (the demon) thus depends. usually 100+ 1000- ? [17:11] depending on order you install services ? [17:13] :/ [17:13] SQLMinUserGID [17:13] this is my problem i think [17:13] clusty i always get Resposta: 550 index.html: Permission denied [17:14] i have that value at 500 [17:14] and in db they are 2001 and 2002 [17:14] uid an dgid 2002 [17:16] got it :p [17:16] files were created by root [17:16] and not the user :S [17:16] :D [17:38] Re: UEC - where is the best place to look to figure out why an instance you just launched went straight from pending to terminated? [17:45] After the patches on karmic today I can at least attempt to launch an instance without getting a 403 error, but they stay at pending for about a minute then go straight to terminated === hggdh|afk is now known as hggdh [18:23] zul: what is the bug number for puppet MIR? [18:24] mathiaz: 408297 [18:24] zul: where did you seeded it? [18:25] server-ship [18:27] zul: where is the MIR for libaugeas-ruby? [18:27] mathiaz: eh? [18:27] zul: it's required by puppet [18:27] zul: http://people.canonical.com/~ubuntu-archive/component-mismatches.txt [18:27] crud [18:28] zul: ^^ search for puppet [18:28] mathiaz: ill get that fixed asap [18:28] zul: same for libshadow-ruby [19:50] anyone seen a postfix crash like this: connect #11 to subsystem private/rewrite: Connection refused ? [19:52] im looking for an irc server app [19:52] to create my own network [19:52] anyone have any ideas? [19:52] apt-cache search ircd ? [19:53] thanks [19:55] KillMeNow: No. Look in the postfix logs. They almost always tell the true story. [20:02] kirkland: i fixed a couple typos in byobu: lp:~jbernard/+junk/byobu [20:08] jbernard: thanks, committed, pushed [20:12] ScottK: i've dug in to the mail.err | mail.warn | mail.log logs and i see when it crashed, but nothing that showed to be the culprit [20:12] hmmm [20:12] got me thinking tho [20:19] mathiaz: both are filed now [21:04] i was spinning through http://people.canonical.com/~ubuntu-archive/NBS, many of these reverse-depend on kvm or qemu which are virtual packages, this is the way we want it to be, correct? === nxvl_ is now known as nxvl [21:06] or should those packages be updated to depend on qemu-kvm? [21:14] jjohansen, you have a minute ? [21:15] zul, might be interested also. [21:44] hello guys - I have a problem which i suspect is udev related. I have a server - administered remotely - running 8.04LTS which has 2 hard disks - physically sda and sdb (well, at least for a long time it was). sdb was a dd exact copy of sda after the server was upgraded to 8.04 from 6.06. Now, after a catastrophic power failure, the server insists on booting from the back-up sdb instead of the main drive sda - like, they have k [21:44] inda swapped places. Any ideas on how to get back to normality? [21:45] mzungu: well there are a few things to look at -- first your bios hard drive boot order [21:45] make sure that the drive you want is first in that order (switch them if you're not sure, and see what happens) [21:45] hmmm - it's remote [21:45] well that's what out-of-band management is for [21:45] or remote hands [21:45] ;) [21:46] but it may also be that your grub config was somehow changed to boot from a different drive than where grub is installed [21:46] are both drives accessible? [21:46] DRAC5++ # Because it lets me do just about anything [21:46] yes - can see both drives [21:47] the grub comments claim it's sda [21:47] but sda seems to have swapped with sdb [21:47] right, that can happen [21:47] which is why ubuntu uses uuids [21:48] not /dev names [21:48] sure [21:49] so you need to establish, first, which drive is being booted by the bios [21:49] so you know which one to edit the grub config of [21:49] and where i might have (unintentionally) messed up was after the upgrade to 8.04, i did a dd if=/dev/sda of=/dev/sdb to make an exact copy [21:49] i suspect the udev identifiers are therefore the same ;) [21:51] mzungu: I don't think they can be the same [21:51] the system which boots is the state after upgrade - some 4 months ago [21:51] afterall uuid stands for unique .. [21:51] which *was* sdb [21:52] udev makes my head spin ;) [21:53] i did try turning off bootable with fdisk - and now i seem to only have sdb and sdc [21:53] and sda is now non-existant! [21:53] (bring back the old days! - bloody hot-plug has a lot to answer for!) [21:55] non-existant where? [21:55] well, i had sda and sdb [21:55] now i have sdb and sdc [21:55] at least, how fdisk sees them [21:56] ok, in fdisk [21:56] which one is being booted off of? [21:56] sda has gone [21:56] sdb i think [21:56] you think? make sure [21:56] (in the fdisk context) [21:56] what? [21:56] mount [21:57] look at where root is being mounted from [21:57] how can i see and relate uuid to a real physical partition? [21:57] there's a /dev map for it [21:57] ok [21:57] smoser: whats up? [21:57] * mzungu goes to check... [21:58] i figured you'd show up just as i was going to leave [21:58] :) [21:58] ls -l /dev/disk/by-uuid/* [21:58] i'm putting comments into bug 431103 that seem to me to indicate changes in kernel cause init (and ohter user-level) boot messages to not go to console [21:58] Launchpad bug 431103 in ec2-init "ssh host key fingerprint no longer available in the console log" [High,Triaged] https://launchpad.net/bugs/431103 [21:59] smoser: okay thanks (late lunch today) [22:02] ah [22:03] ok - that's showing the sdc drive [22:04] lrwxrwxrwx 1 root root 10 2009-09-22 23:24 /dev/disk/by-uuid/ed3d5dd5-7c95-4e97-86eb-60f66cf54f4b -> ../../sdc2 [22:04] sdc2 being the root partition [22:08] how do i con it into using what it now thinks is sdb? [22:25] jjohansen, i added comment and console-output.tar.gz to that bug. [22:25] please read, and refute me if you think i'm wrong [22:25] i've got to run [22:28] smoser: okay