| vlad_starkov | TJ-: OK, "(initramfs)" appears :) | 00:00 |
|---|---|---|
| TJ- | vlad_starkov: OK, so that confirms that its something caused during the upstart init process... so its a case of figuring that out now. First, lets try booting to a bash shell instead of starting upstart. Remove "break=init" and add "init=/bin/bash" and reboot. If that works, you've booted into the real root file-system but used BASH as the init process... which'll not do any configuration | 00:02 |
| vlad_starkov | TJ-: OK, I'll try it now... | 00:04 |
| vlad_starkov | TJ-: it doesn't freeze now, but it stops with blinking cursor, and it doesn't ask for crypt passphrase as usual | 00:07 |
| TJ- | vlad_starkov: Aha... well that makes sense! | 00:07 |
| TJ- | vlad_starkov: OK... you can lose the "init=/bin/bash" since with using cryptsetup the init rules need to run to unlock the volume | 00:08 |
| vlad_starkov | TJ-: yepp | 00:08 |
| vlad_starkov | TJ-: is there any option? | 00:12 |
| TJ- | vlad_starkov: None I can think of right now... having encryption complicates the debug process immensely. Is there a spare (bootable) disk in there you could install an unencrypted testing environment into? | 00:15 |
| vlad_starkov | TJ-: this server is empty, I can reinstall the system on it | 00:15 |
| TJ- | vlad_starkov: without hands/eyes-on its very hard to work with these kinds of issues... often I spot subtle clues which isn't easy to do with IRC in the way | 00:16 |
| vlad_starkov | TJ-: You're right | 00:18 |
| vlad_starkov | TJ-: so could you point me to the right way, what strategy should I apply? | 00:19 |
| vlad_starkov | TJ-: As I see now, the point is to reinstall system without encryption. Then try to boot, if it still doesn't boot, try "init=/bin/bash". If it boots this way, try to make something in /sbin/init | 00:21 |
| TJ- | vlad_starkov: My next step would be to create a minimal command-line environment installation on a USB flash storage device (I actually carry one around with me for this kind of situation) that would then allow me to selective disable kernel modules that I can see being loaded, until I find the module(s) causing the lockups | 00:21 |
| vlad_starkov | TJ-: sounds awesome. Is there a fast and easy way to make it? (some manual you could recommend) | 00:22 |
| TJ- | vlad_starkov: If you can get into /bin/bash, you're in the read-write root file-system, and can then edit files in /etc/modprobe.d/ to blacklist modules that you suspect are to blame for the lockups. It's an iterative process using guesswork, hunches, and clues from the logs in /var/log/ .. especially /var/log/dmesg and /var/log/kern.log and /var/log/udev | 00:23 |
| vlad_starkov | TJ-: I understand | 00:23 |
| TJ- | vlad_starkov: I've never come across one; this kind of sys-admin stuff comes from lots of experience with similar problems which develops an intuition about it | 00:24 |
| vlad_starkov | TJ-: Which distro do you recommend for usb? | 00:27 |
| TJ- | vlad_starkov: Ubuntu :) Mine starts from an ubuntu-minimal install and then has had all my preferred tools added. You can do something similar by installing from the Ubuntu server ISO to a USB device and not selecting any server packages, and then customising it after it has managed to boot (assuming it can!) | 00:29 |
| vlad_starkov | TJ-: Am I right thinking that I can do the same with Ubuntu installer? | 00:31 |
| TJ- | vlad_starkov: The server installer? That was the one I was referring to. I don't think it can be done automatically from the LiveCD installer. | 00:33 |
| vlad_starkov | TJ-: I think, this is what I need https://wiki.ubuntu.com/LiveUsbPendrivePersistent | 00:39 |
| TJ- | vlad_starkov: I'd guess, being a LiveCD image, it'll hang like the liveCD itself does. That's why these situations need a really minimal non-GUI install | 00:40 |
| TJ- | vlad_starkov: What you need is something that you can easily change the boot process on and will do the absolute minimum necessary to get to a working terminal | 00:41 |
| vlad_starkov | TJ-: I understand. So don't you know if there "ready to burn" clone of working minimal ubuntu that I can download and dd to my usb flash? | 00:43 |
| vlad_starkov | TJ-: for the moment all I found is how to install Linux on USB | 00:44 |
| TJ- | It's so long since I needed that I don't know; I've always created my own | 00:44 |
| vlad_starkov | TJ-: Is it possible to make it with 1 MacBook and 1 USB flash stick? | 00:44 |
| TJ- | vlad_starkov: a USB device is the same as any other disk as far as installing goes. You only need 'special' measures for USB if creating a clone of a LiveCD ISO image | 00:45 |
| vlad_starkov | TJ-: as I understood I need kinda virtualization environment (program like VirtualBox) in which I should run Ubuntu Server installer and then I have to choose USB stick as a hard drive on which the installation should be done | 00:49 |
| TJ- | vlad_starkov: That sounds about right, yes | 00:51 |
| vlad_starkov | TJ-: do you know if the Parallels is suitable for it? | 00:54 |
| TJ- | I'd assume any hypervisor should be OK for doing that | 00:55 |
| vlad_starkov | TJ-: Ok, I'm in. | 01:15 |
| vlad_starkov | TJ-: ls /etc/modprobe.d: | 01:15 |
| vlad_starkov | TJ-: there are only blacklist files | 01:16 |
| TJ- | vlad_starkov: You mean you've booted the server successfully to a /bin/bash prompt? | 01:17 |
| vlad_starkov | TJ-: ooops, that's my fault, I understood that I do incorrectly just a second ago | 01:19 |
| vlad_starkov | TJ-: How to know what kernel modules are being loaded on boot? | 01:20 |
| TJ- | vlad_starkov: By reading the previous boot's log-files in /var/log/ ... the previous boot(s) will have log-files with names including a "0" or "1", especially dmesg and kern.log | 01:25 |
| vlad_starkov | TJ-: in my syslog I have multiple *BAD*gran_size messages, is it normal? | 01:25 |
| TJ- | vlad_starkov: I don't recognise that message and no it doesn't sound normal | 01:26 |
| vlad_starkov | TJ-: there are only media-info and syslog in /var/log | 01:26 |
| TJ- | vlad_starkov: I think I'd start by looking at all the PCI hardware recognised by "lspci -nn" ... then I'd look-up the PCI device_vendor IDs of each device and locate the kernel module that drives it, and disable them one at a time by adding their names to a blacklist file in /etc/modprobe.d/ before booting normally (but with "debug") and seeing if the soft lockups go away | 01:27 |
| vlad_starkov | TJ-: Found in /var/log/upstart/module-init-tools.log: "FATAL: Module rtc not found." | 01:42 |
| JanC | some people/discussions with/about that error mention booting with the 'disable_mtrr_cleanup' kernel parameter as a workaround? | 01:46 |
| JanC | also, sometimes it helps to change the amount of RAM available to an on-board GPU to the minimum? | 01:51 |
| TJ- | vlad_starkov: I have to go to bed now, but I've written a compact BASH scriptlet that will determine what loadable kernel modules are required just by using the output of lspci, which you can run from the "init=/bin/bash" terminal without having to run upstart. http://paste.ubuntu.com/6648944/ | 02:31 |
| vlad_starkov | TJ-: Thanks! You helped me so much today | 02:32 |
| vlad_starkov | TJ-: I appreciate it! Thank you for your time | 02:33 |
| TJ- | vlad_starkov: The output lists, for each device, any loadable kernel modules that can manage the device. Those are the module names you should add to blacklist entries in "/etc/modprobe.d/blacklist-testing.conf" in the form of one per line "blacklist <module_name>" | 02:33 |
| Logan_ | xnox: tsp tsp, synced over your own autoreconf delta, and now it's not building on ppc64el: https://launchpad.net/ubuntu/trusty/+source/ntrack/016-1.2 Want me to reintroduce it? | 02:34 |
| Logan_ | *tsk tsk lol | 02:34 |
| TJ- | vlad_starkov: This is example output, from my laptop: http://paste.ubuntu.com/6648963/ | 02:34 |
| === henrix_ is now known as henrix | ||
| xnox | Logan_: did I? *sigh* yes, please. | 02:34 |
| Logan_ | doing | 02:34 |
| vlad_starkov | TJ-: OK | 02:35 |
| vlad_starkov | TJ-: will try it tomorrow | 02:35 |
| TJ- | vlad_starkov: Good luck with it | 02:35 |
| vlad_starkov | TJ-: Thanks | 02:38 |
| TJ- | vlad_starkov: Be best to continue support in #ubuntu-server rather than this channel, too; this is for Ubuntu development not user support. I'm in that channel too | 02:39 |
| vlad_starkov | TJ-: I'll find you there. Would be glad to continue discussing this problem later on | 02:40 |
| === brainwash_ is now known as brainwash | ||
| uBUXUBu | odd behaviour in ubunut software center-crashed-report sent-after crash software center acting strange-when i tried to read about software packages they were all blanked out and had lines drawn thru them as though someone had taken a pencil and scrathced things out? very bizarre-rebooted all normal? | 08:01 |
| uBUXUBu | ubuntu* | 08:02 |
| uBUXUBu | odd cause in using 12.04 LTS | 08:02 |
| darkxst | uBUXUBu, file a bug with a screenshot | 08:09 |
| uBUXUBu | well as i just said i rebooted and its gone | 08:10 |
| uBUXUBu | but i did file report | 08:11 |
| ogra_` | ogra 2038 4.6 53.6 8997504 4076544 ? Ssl Dez11 1119:24 /usr/lib/x86_64-linux-gnu/hud/hud-service | 09:40 |
| ogra_` | hmm | 09:40 |
| ogra_` | 4G for the hud ... | 09:40 |
| infinity | adconrad 2559 0.0 0.1 372556 29996 ? Ssl Dec21 4:14 /usr/lib/x86_64-linux-gnu/hud/hud-service | 09:43 |
| infinity | I think you've done something special to yours. | 09:43 |
| ogra_` | well, not that i know of ... i use it ... | 09:43 |
| ogra_` | i wonder fi it caches all requests ever made in a session | 09:43 |
| ogra_` | i suspect a reboot is in order ... | 09:46 |
| === Chipzz_ is now known as Chipzz | ||
| * ogra_` glares at update-manager ... | 10:21 | |
| ogra_` | why the heck do i have texlive-latex-extra-doc installed ... and why is it 300MB bit !! | 10:22 |
| ogra_` | *big | 10:22 |
| ogra_` | argh | 10:41 |
| ogra_` | i think this upgrade killed my libcd | 10:42 |
| ogra_` | *libc | 10:42 |
| * ogra_` gets a "file not found" for everything | 10:42 | |
| ogra_` | sigh | 10:43 |
| cjwatson | ogra_`: If you had libc6-amd64:i386 installed, see the instructions near the top of https://wiki.ubuntu.com/Touch/Emulator | 10:43 |
| ogra_` | i surely had, and usually make sure it doesnt get uninstalled | 10:43 |
| ogra_` | seems soemthing removed it | 10:43 |
| cjwatson | The upgrade from 2.17 to 2.18 may have had similar effects to uninstallation | 10:44 |
| ogra_` | ah | 10:44 |
| cjwatson | (Perhaps, I don't know) | 10:44 |
| ogra_` | hmm | 10:46 |
| ogra_` | i still have 2.17 installed | 10:46 |
| ogra_` | sigh, and i cant sudo indeed | 10:49 |
| ogra_` | sigh | 11:27 |
| * ogra_` fixes the instructions on the wiki ... linking to the linker of the initrd kind of doesnt really work :P | 11:28 | |
| ogra_` | cjwatson, thanks ! | 11:28 |
| ogra_` | (all back up again) | 11:28 |
| xnox | How to do a fakesync, when debian's dfsg'ed upstream version number is lower than ubuntu's dfsg'ed upstream version number? | 13:32 |
| xnox | opencolorio 1.0.8~dfsg0-2.1 (debian), 1.0.8+repack1-0ubuntu3 (ubuntu) | 13:33 |
| * xnox is thinking to rename debian's tarball into 1.0.8+repack2 and write a sensible explanatory changelog entry that this is in-fact a fake-sync. | 13:34 | |
| === sraue_ is now known as sraue | ||
| === yofel_ is now known as yofel | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!