=== dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates === gtaylor2 is now known as gtaylor === txwikinger2 is now known as txwikinger === the-internet is now known as fw0rd === dendrobates is now known as dendro-afk === orangeac1e is now known as orangeacme === daker_ is now known as daker === Hussain is now known as Guest6540 [12:17] kim0: thanks for your support of my membership application :) [12:17] TeTeT: most welcome ;) [12:21] membership eek [12:21] debian images away! [12:22] http://www.rightscale.com/library/multi_cloud_images/All?search%5Badvanced_search%5D=&search%5Bfilter_value%5D=debian&search%5Bfilter_type%5D=title&search%5Bprice%5D=&search%5Border%5D=date_desc&x=0&y=0 === daker_ is now known as daker === kdoandal is now known as nevrax === dendro-afk is now known as dendrobates === uvirtbot` is now known as uvirtbot === dendrobates is now known as dendro-afk [15:57] TeTeT: me again! ... network and disk do not appear to be an issue with my performance. I have been trying to find a way to run a GFlop test on my instance. no luck so far. [15:59] i did notice something interesting in my cpuinfo... model name : QEMU Virtual CPU version 0.12.5 [15:59] yesterday you asked about QEMU === dendro-afk is now known as dendrobates [16:23] what does it mean if I find this message in my libvirt logs... Could not initialize KVM, will disable KVM support [16:24] My instances run, however slow. [16:24] hi navanjr , I think it means you're running with qemu instead of kvm. Are you sure that kvm is enabled in the BIOS? [16:24] i thought i was sure... [16:24] navanjr: what's the output of kvm-ok on the NC [16:25] nate@nc:/var/log/libvirt/qemu$ kvm-ok, INFO: Your CPU supports KVM extensions, INFO: /dev/kvm exists, KVM acceleration can be used [16:25] kim0: I had it happen once in an on site training that kvm-ok reported fine, but still qemu was used [16:26] :s [16:30] the line right above... open /dev/kvm: Permission denied [16:35] navanjr: on the node, if you just start kvm on the command line, what's the output? [16:36] navanjr: if there's none, means you're good [16:37] well, im not good. :) [16:38] open /dev/kvm: Permission denied, Could not initialize KVM, will disable KVM support, pci_add_option_rom: failed to find romfile "pxe-rtl8139.bin", Could not initialize SDL - exiting [16:38] do i just need to chown of the /dev/kvm ? [16:40] nate@nc:/dev$ ls -lh kvm, crw-rw---- 1 root root 10, 232 2011-03-18 11:30 kvm [16:41] navanjr: it's root:kvm for me [16:41] on natty [16:42] ok i changed it.... [16:42] kim0, navanjr : root:kvm for me on Lucid too [16:43] so when i "sudo kvm" on the node [16:43] here is what i get... [16:43] pci_add_option_rom: failed to find romfile "pxe-rtl8139.bin" [16:43] init kbd. [16:43] Could not open SDL display [16:44] UEC shouldn't really use kvm's SDL interface [16:44] any worries there? [16:44] navanjr: install kvm-pxe to let the romfile problem go away [16:44] navanjr: looks better than before I'd say [16:44] navanjr: maybe time to launch another instance and see if it's any faster [16:44] cool, i really appreciate you help so far! [16:45] ill report back [16:50] be back later === herki_ is now known as herki [17:07] smoser: re the udev thing.. I notice now that udev has moved from 165 to 167 since you reported the Connection Refused bug. Are you still seeing problems in the automated tests? [17:11] thanks guys! my instances are now running in KVM. I have some new issues now. Now my console says there are errors in clould-init. all my new test are without any user-data. [17:12] last line on the console... landscape-client is not configured, please run landscape-config. [17:12] navanjr: r u using the default networking mode, managed-novlan [17:12] managed-novlan [17:12] navanjr: that one is normal .. ignore it :) [17:13] the instance is running... virsh list says so [17:13] responds to ping [17:13] but cant ssh [17:13] It's probably hung waiting for user-data [17:13] takes like 20 mins to time out ! [17:13] oh [17:14] consuming user data failed! [17:14] Traceback (most recent call last): [17:14] File "/usr/bin/cloud-init", line 90, in [17:14] main() [17:14] I'm not really sure why the user-data service would have problems in uec [17:17] here is my console output... (fwiw) http://pastebin.com/QcJgwFgS === cmagina is now known as cmagina-lunch [17:26] navanjr: got better performance now? [17:27] sure Do! [17:27] having some trouble with cloud-init [17:27] http://pastebin.com/QcJgwFgS [17:33] navanjr: hmm, that's bad, can you try a more recent image, like the released maverick one and see if that works? If not it might mean that your front-end is not setup correct [17:35] it was all working previously to the chmod [17:36] and the apt-get install kvm-pxe [17:36] on the NC [17:36] navanjr: maybe a temporary issue, launch another instance and check [17:37] navanjr: I believe that very old lucid images had a problem with timeouts and cloud-init, but it was fixed, so I asked for trying out maverick [17:40] ok will do! ill let you know. thanks, TeTeT === cmagina-lunch is now known as cmagina [18:00] SpamapS, i believe we saw the bug in beta1 testing [18:01] smoser: hrm [18:01] http://jenkins.qa.ubuntu-uk.org/view/natty-ec2/job/natty_ec2_ap-southeast-1_all-zones/3/artifact/ap-southeast-1b/i386/m1.small/instance-store/i-c518d190/3b5d50de-fc22-4e17-88fd-2b3a74e8e63b-timeout.console.txt/*view*/ [18:01] that is a beta1 log [18:02] smoser: its *really* a bitch to debug udev [18:02] now, beta1 was udev 166-0ubuntu6 [18:02] SpamapS, thats why i'm glad i got you to do it [18:03] udevd[217]: bind failed: Address already in use [18:03] that, I think, is relaly the crux of the problem [18:03] does initramfs run a udev? [18:04] it might. [18:04] * SpamapS can answer that himself [18:06] SpamapS, [18:06] $ lsinitramfs /boot/initrd.img-$(uname -r) | grep udev [18:07] says the answer to your question is "yes" [18:08] from logs i can't be certain if those error messages are before or after transition to "the real init" [18:10] diff from 166-0ubuntu6 -> 167-0ubuntu1 is [18:10] 54 files changed, 509 insertions(+), 1003 deletions(-) [18:13] SpamapS, http://paste.ubuntu.com/590359/ is the "summary of changes" [18:16] interesting.. udevd writes to /dev/kmsg .. didn't know that but it makes sense [18:16] [ 0.630017] udev[59]: starting version 167 [18:16] that explains that [18:18] smoser: this one MIGHT be related: udevd: initialize fds, for proper close() on exit [18:18] smoser: possible that the initramfs udevd wasn't closing its sockets properly leading to them being left open.. but I don't think AF_LOCAL sockets do that. [18:19] yeah, that entry stuck out a bit [18:20] smoser: do you guys run constant tests like these? 167 has been in for a few days now [18:20] * SpamapS hopes nobody picks up on the desperation in his tone :) [18:20] we're not to the point of running daily tests [18:20] we're getting there, thanks to jamespage === daker is now known as daker_ [18:22] SpamapS, http://paste.ubuntu.com/590361/ is the message you were looking for [18:22] thats the upstream commit [18:23] *hm* [18:24] if thats the case then its very hard to trigger the race artificially.. I have to get udevd to start sooner [18:29] smoser: been a long time since I last worked with something this complicated in bash, but finally remembered a few things and learned others. This is what I'm doing so far, but it is just testing as since I started working on it in the past couple hours. Does this look sane to you so far? http://pastebin.ubuntu.com/590364/ [18:34] regions_array=(${regions#,}) [18:34] regions has a leading , ? [18:35] smoser: yes, kernels too and presumming that initrds too [18:35] ah, yeah it does. [18:35] you should fix that issue up aabove though. [18:35] smoser: ok [18:36] just because it looks confusing [18:36] acutally, its probably better if you actually just create the array above [18:37] then you dont have to mess with splitting it back up [18:37] smoser: ok cool [18:38] youll have to make a temporary output file [18:38] --output "${output}" [18:38] if you give them all the same they'll probably truncate each others [18:38] ok cool [18:38] * RoAkSoAx takes notes [18:39] the one other thing, is that you're inside an arch loop there [18:39] so right now you'd be publishing all regions at the same time for i386 [18:39] and then for amd64 [18:39] smoser: yeah I'm looking into handling that differently [18:39] k [18:39] thanks , RoAkSoAx [18:40] smoser: i was thinking on something like waiting for the PIDS so, remove the wait after the for, and put it right after the arch while [18:41] and queue all the PIDs, and wait for the queue to be empty [18:42] eyah. you might be able to use something like gnu parallel for that. [18:42] ie, you could just collect all the things up that you wan to run and pass it off to something else. [18:49] ok will take a look at that [19:16] smoser: I just got the "Connection refused" bug on a vm w/ udev 167 [19:17] actually no! [19:17] it has 166 [19:17] forgot to upgrade this one [19:17] damnit [19:17] well, good news for local recreate [19:17] you can launch kvm with -nographic [19:17] in a loop [19:17] yeah its happened once before [19:17] and just kill it after 30 seconds or whatever [19:18] well now this is interesting.. [19:18] the udev from kmsg says version 167.. [19:18] but dpkg shows 166 [19:19] err.. udevd is 167 but dpkg -l is 166. Ok well now I've gone cross eyed === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [21:18] whats the trick to publishing maverick image? [21:19] nate@nateDesktop:~$ uec-publish-tarball $UEC_IMG.tar.gz $RELEASE-$TIMESTAMP [21:19] Wed Apr 6 15:11:29 CDT 2011: ====== extracting image ====== [21:19] Warning: no ramdisk found, assuming '--ramdisk none' [21:19] kernel : maverick-server-uec-amd64-vmlinuz-virtual [21:19] ramdisk: none [21:19] image : maverick-server-uec-amd64.img [21:19] Wed Apr 6 15:11:36 CDT 2011: ====== bundle/upload kernel ====== [21:19] failed to register maverick-server-uec-amd64-vmlinuz-virtual.manifest.xml [21:19] failed: euca-register --name maverick-server-uec-amd64-vmlinuz-virtual maverick-20110406151031/maverick-server-uec-amd64-vmlinuz-virtual.manifest.xml [21:19] Image: Image registration failed because the manifest referenced is invalid or unavailable.failed to upload kernel === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates === niemeyer is now known as niemeyer_away === dendrobates is now known as dendro-afk === Nolar is now known as LawsuitBot === LawsuitBot is now known as Nolar