| verwilst (~verwilst@78-21-41-88.access.telenet.be) joined #fai. |
| verwilst (~verwilst@78-21-41-88.access.telenet.be) left irc: Quit: Ex-Chat |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. |
| kai_bo (~kai@statdsl-085-016-072-173.ewe-ip-backbone.de) joined #fai. |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Ping timeout: 480 seconds |
| Nic0 (~courtel@fw.tls.cena.fr) joined #fai. |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) joined #fai. |
| ErKa (keryell@m6.wifi.enstb.org) joined #fai. |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving |
| Mrfai (~lange@suenner.informatik.uni-koeln.de) joined #fai. |
| 08:42 FBI | fai: 3 lange committed revision 5404 to 7 fai: make release |
| ErKa (keryell@m6.wifi.enstb.org) left irc: Ping timeout: 480 seconds |
| 09:09 FBI | fai: 3 lange committed revision 5405 to 7 fai: tag release 3.2.20 |
| verwilst (~verwilst@92.48.144.162) joined #fai. |
| stockholm (~andreas@marta.schuldei.org) joined #fai. |
| verwilst (~verwilst@92.48.144.162) left irc: Read error: Connection reset by peer |
| 10:42 mikap | Mrfai: Calling debootstrap sid --arch i386 --exclude=dhcp-client,info /grml-live/grml-broken/grml_chroot http://debian.netcologne.de/debian/ => that's a cmdline that doesn't work, should be debootstrap --arch i386 --exclude=dhcp-client,info sid /grml-live/grml-broken/grml_chroot http://debian.netcologne.de/debian/ so it works, and return code is 0 |
| kai_bo (~kai@statdsl-085-016-072-173.ewe-ip-backbone.de) left irc: Quit: Verlassend |
| kai_bo (~kai@statdsl-085-016-072-173.ewe-ip-backbone.de) joined #fai. |
| 11:33 Mrfai | mikap: which fai version? IIRC a former version reported the option in a different order than they were actually used during calling debootstrap |
| 11:33 mikap | Mrfai: the most recent |
| 11:34 mikap | ah, no not 3.2.20 but 3.2.19 - will retry, give me a second |
| 11:34 Mrfai | 3.2.20 has no changes in make-fai-nfsroot |
| 11:36 Mrfai | this is from make-fai-nfsroot: |
| 11:36 Mrfai | echo "Calling debootstrap $1 $NFSROOT $2" |
| 11:36 Mrfai | LC_ALL=C call_verbose debootstrap $FAI_DEBOOTSTRAP_OPTS $1 $NFSROOT $2 |
| 11:37 Mrfai | looks fine to me |
| 11:37 Mrfai | $1 is sid in your case, and $2 is the mirror url |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) left irc: Quit: Leaving. |
| atlantide (~antares@mail.lipn.univ-paris13.fr) left irc: Quit: Quitte |
| verwilst (~verwilst@router.begen1.office.netnoc.eu) joined #fai. |
| ErKa (keryell@m6.wifi.enstb.org) joined #fai. |
| proti (~seb@calypso.frankengul.org) joined #fai. |
| 13:21 proti | morning |
| 13:23 Mrfai | hi |
| 13:26 proti | I wondering how the nfs root is mounted on the client ? |
| 13:29 Mrfai | the initrd does this for you via NFS |
| 13:31 proti | initrd gets the nfsroot/live/filesystem.dir mounted on /live. |
| 13:32 proti | Making the nfsroot, the client says that it cannot mount /var/lib/fai/config on nfs. |
| 13:32 proti | Is there something like pivot_root called by initrd ? |
| 13:35 proti | Maybe I should start at the beginning. |
| 13:36 proti | I'm trying to update fai from 3.2.8 to 3.2.20. |
| 13:36 Mrfai | /var/lib/fai/config is the config space. This is also mounted via NFS. On the server side this is /srv/fai/config |
| 13:36 Mrfai | check your exports file |
| 13:36 proti | I did the make-fai-nfsroot to update the whole thing. |
| 13:36 proti | Then I tried a sysinfo. |
| 13:36 proti | The mount fails because the /var/lib/config does not exists. |
| 13:37 proti | /var/lib/fai/config* |
| 13:39 proti | I created one then the install failed at instsoft with lots errors (/var/lib/dpkg/status not found). |
| 13:39 proti | So I supposed I got it wrong somewhere before. |
| 13:40 proti | I felt that the /var is not mounted correctly. |
| 13:41 proti | If I understood well the client mounts nfsroot then do an union mount and use this filesystem to run fai. |
| 13:43 Mrfai | yep, but it seems that you have problems with the config space, not the nfsroot |
| 13:44 proti | I created the /var/lib/fai/config and then, the mount was ok (sysinfo, and so on). |
| 13:45 proti | But then the build failed in a miserable way (at instsoft). |
| 13:46 Mrfai | put the fai.log onto paste.debian.net |
| 13:51 proti | By the way I'm on ubuntu. |
| 13:51 proti | http://paste.debian.net/35197/ |
| 13:52 proti | The /var/lib contains only one directory 'discovery'. |
| 13:52 proti | bigboss is the server and marseille is the client |
| 13:54 Mrfai | you nfsroot is fine, otherwise you won't see the red logo (the copyright message) |
| 13:54 proti | So why don't I see any of the /var/ nfsroot (dpkg apt aptitude and so on ) ... |
| 13:54 Mrfai | check why /var/lib/fai/config does not exist. Is it inside a ramdisk/tmpfs? |
| 13:55 proti | Yes initrd. |
| 13:55 Mrfai | no at this point the initrd is not used any more |
| 13:55 proti | Where are we then on the nfsroot + unionfs ? |
| 13:56 Mrfai | yep |
| 13:56 proti | Maybe not, because there is much more in the /var/lib of the nfsroot. |
| 13:57 proti | And I don't see any repository but the discover. |
| 13:57 proti | directory* |
| 13:57 proti | On the nfsroot, there is a /var/lib/fai/config. |
| 13:58 Mrfai | what does df /var or mount|grep var say on the install client? |
| 14:00 proti | mount aaufs on / type aufs (rw,noatime,relatime,xino=/live/cow/.aufs.xino,br:/live/cow=rw:/live/image/live/filesystem.dir=ro) |
| 14:00 proti | There is no /var mounted |
| 14:01 Mrfai | ok. Wast does /var contain? ls-lR or find -ls (doe not paste here, if too long) |
| 14:02 proti | http://paste.debian.net/35201/ |
| 14:04 Mrfai | let me check here |
| 14:05 Mrfai | on my install server I have a directory /srv/fai/nfsroot/live/filesystem.dir/var/lib/fai/config/ |
| 14:05 Mrfai | this is also avaialable on the install client |
| 14:06 Mrfai | do you also have this dir on the install server? |
| 14:07 proti | yes I have. |
| 14:07 proti | But not on the client |
| 14:08 Mrfai | strange |
| 14:09 proti | Is there a cleanup of /var done ? Because I can see much more files on the nfsroot /var than there is on the client. |
| 14:10 Mrfai | I would blame Ubuntu (maybe upstart) that does something different with /var. Maybe ubuntu copies /var from the initrd and uses it, instead of the /var from inside the nfsroot. |
| 14:11 proti | Let me check.... |
| 14:11 Mrfai | A fix would be, to mkdir -p $FAI in get-config-dir or get-config-dir-nfs |
| 14:11 proti | this is what I did previously. |
| 14:13 proti | But then the install_packages fails at installing packages. |
| 14:13 Mrfai | There may be more directories missing |
| 14:13 proti | Indeed, apt, aptitude, dpkg in var/lib. |
| 14:14 proti | But this makes no sense. |
| 14:14 proti | is a union mount of filesystem.dir and tmpfs. Why /var/ is different ? |
| 14:16 Mrfai | I'm sure its not /var from the nfsroot. Check this: mkdir var/this-is-a-test-directory inside the nfsroot on the install server. Then reboot the client and see if this directory is available |
| 14:16 Mrfai | if not, blame Ubuntu for its behaviour ;-) |
| 14:23 proti | 3.2.8 was ok same distrib (hardy). |
| 14:24 proti | Hum, I touched /var/test-fairoot |
| 14:24 proti | And then, the file is there on /var of the client (no reboot needed). |
| 14:25 proti | So this is the right /var. Seems that all other dir are deleted on the tmpfs union mount. |
| 14:26 Mrfai | your ls -lR (paste 35201) was made on the install server, but not inside the nfsroot |
| 14:26 Mrfai | check again the /var inside the nfsroot. Is there a var/lib/fai directory |
| 14:26 proti | No. |
| 14:27 proti | It was the client. |
| 14:27 Mrfai | sorry, my fault |
| 14:28 Mrfai | try to create a var/lib/test-... directory. Maybe only /var/lib is affected |
| 14:32 proti | all the test-fairoot files I touched are seen in the client. I'm rebooting the client to see if see them after rebooting |
| 14:38 proti | Ok, only lib is impacted. |
| 14:40 proti | I'm trying to break=init to see if the files are there or not. |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) joined #fai. |
| 14:52 mikap | Mrfai: I'm talking about dirinstall ;) |
| 14:52 mikap | Mrfai: # dpkg -L fai-client | xargs grep FAI_DEBOOTSTRAP_OPTS |
| 14:52 mikap | /usr/lib/fai/subroutines-linux: echo "Calling debootstrap $1 $FAI_DEBOOTSTRAP_OPTS $FAI_ROOT $2" |
| 14:52 mikap | /usr/lib/fai/subroutines-linux: yes '' | LC_ALL=C debootstrap $FAI_DEBOOTSTRAP_OPTS $1 $FAI_ROOT $2 |
| 14:52 mikap | definitely not the same cmdline layout |
| 14:55 Mrfai | yep, this seems to be a bug |
| 14:58 Mrfai | but isn't the yes ''| pipe correct? It's only the echo which is wrong |
| 15:02 mikap | Mrfai: http://paste.pocoo.org/show/U6YibYVfRskOf1SaWtb3/ |
| 15:02 mikap | Mrfai: # debootstrap sid --arch i386 --exclude=dhcp-client,info /grml-live/grml-broken/grml_chroot http://debian.netcologne.de/debian/ |
| 15:02 mikap | E: No such script: --exclude=dhcp-client,info |
| 15:02 mikap | => definitely a broken cmdline |
| 15:02 mikap | the 'sid' should be behind the options |
| 15:03 mikap | the fix for this is trivial |
| 15:03 mikap | though my dirinstall problem itself is still present |
| 15:05 Mrfai | IMO the debootstrap call finished without a problem. Try debootstrap, without fai. Then check the exit code. |
| 15:05 Mrfai | Someone told me that sometimes debootstrap is succsessful, but has exit code 141 |
| 15:05 mikap | Mrfai: http://grml.org/tmp/0001-Fix-order-of-debootstrap-cmdline-for-working-copy-pa.patch |
| 15:06 mikap | Mrfai: this patch addresses the broken log entry of debootstrap |
| 15:06 mikap | Mrfai: already tested and report it: |
| 15:06 mikap | # debootstrap --arch i386 --exclude=dhcp-client,info sid /grml-live/grml-broken/grml_chroot http://debian.netcologne.de/debian/ 1>/dev/null ; echo $? |
| 15:06 mikap | 0 |
| 15:06 mikap | => definitely OK |
| 15:07 Mrfai | ok. |
| 15:07 Mrfai | The message: Calling debootsrtap..... has a different order of options than the call itself |
| 15:08 mikap | Mrfai: yeah, that's what my patch addresses |
| 15:08 mikap | now investigating on the b0rken dirinstall of FAI > 3.2.17 |
| 15:09 Mrfai | ok, I will apply the patch |
| 15:11 FBI | fai: 3 lange committed revision 5406 to 7 fai: subroutines-linux: fix log message |
| 15:13 Mrfai | mikap: add a set -x in the subroutine, to see what going on there |
| 15:13 mikap | Mrfai: doing that already ;) |
| 15:13 mikap | Mrfai: wrong changelog ;) |
| verwilst (~verwilst@router.begen1.office.netnoc.eu) left irc: Ping timeout: 480 seconds |
| 15:20 Mrfai | mikap: I use debootstrap 1.0.10lenny1 on my etch install server. No problems. |
| 15:21 mikap | Mrfai: amd64 build system? |
| 15:22 Mrfai | it's an amd64 kernel with 32 bit user space tools. I only tested to dirinstall a 32 bit system. Building a 64bit nfsroot (with make-fai-nfsroot) is working |
| 15:22 mikap | hm |
| 15:23 mikap | well, it's definitely just FAI -> old fai: works; upgrade to recent fai (no further changes): b0rken |
| 15:23 mikap | + RC=141 |
| 15:23 mikap | + echo 'task_error 801 141' |
| 15:23 mikap | task_error 801 141 |
| 15:23 Mrfai | 141, that's it |
| 15:23 mikap | WTF does fai get that 141 from? |
| 15:23 mikap | it's *not* debootstrap |
| 15:24 Mrfai | from an email: |
| 15:24 Mrfai | debootstrap completes with the following message: |
| 15:24 Mrfai | "I: Base system installed successfully." |
| 15:24 Mrfai | but the return value is 141 not 0. task_error trips on this. |
| 15:24 Mrfai | I'm not sure who is to blame. Couldn't find any description what |
| 15:24 Mrfai | returncode 141 is supposed to mean so far. |
| 15:25 Mrfai | have a look at |
| 15:25 Mrfai | https://lists.uni-koeln.de/pipermail/linux-fai/2008-November/006345.html |
| 15:25 Mrfai | and the following message |
| 15:26 mikap | Mrfai: it's just FAI which changes the behaviour ;) |
| 15:26 Mrfai | yep, but FAI assumes that the return code is 0 if everything is ok. |
| 15:26 Mrfai | can you upgrade your debootstrap? |
| 15:27 mikap | Mrfai: will do, but this doesn't fix the problem but just the symptom |
| 15:28 Mrfai | In the past FAI didn't correctly used the return code when calling task_error. This was fixed, but now is causes an error, because debootstrap has a wrong exit code IMO. |
| 15:28 mikap | I: Base system installed successfully. |
| 15:28 mikap | debug: RC = 0 |
| 15:28 mikap | looks like the 'yes' is involved?! |
| 15:29 Mrfai | IMO the yes can be removed, because debootstrap never prompted for input in my environment |
| 15:30 mikap | ACK |
| Action: Mrfai switching to screen, driving home |
| Mrfai (~lange@suenner.informatik.uni-koeln.de) left irc: Quit: leaving |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. |
| 15:33 mikap | Mrfai: https://lists.uni-koeln.de/pipermail/linux-fai/attachments/20081112/ff052c82/attachment.bin would be an approach if the problem can't be fixed otherwise, though still investigating |
| 15:33 mikap | look: |
| 15:33 mikap | I: Base system installed successfully. |
| 15:33 mikap | Debug: RC=0 |
| 15:33 mikap | Error in task extrbase. Traceback: task_error task_extrbase task task_dirinstall task task_action task main |
| 15:33 mikap | with: |
| 15:33 mikap | yes '' | LC_ALL=C debootstrap $FAI_DEBOOTSTRAP_OPTS $1 $FAI_ROOT $2 |
| 15:33 mikap | RC=$? |
| 15:33 mikap | echo "Debug: RC=$?" |
| 15:33 mikap | return $RC |
| 15:39 proti | Mrfai: The mount is okay now. It wasn't working before and then it works now. And I still don't known why :-( |
| 15:41 proti | Must be race condition, or voodo magic. I don't known. I added a few [ -d /var/lib/fai ] && echo Directory is there #. |
| 15:41 proti | And suddendly, it has found all of the directory of /var. |
| 15:42 proti | Which remembers me that I had the same problem with 3.2.8. |
| 15:42 proti | Go figure... |
| 15:45 mikap | Mrfai: no idea why: but as soon as I remove the yes-pipe it works |
| 15:50 mikap | Mrfai: what should yes '' do?! did you mean just "yes"? |
| 15:50 mikap | in sense of: what should the yes call with '' do overall? (besides that the yes is definitely not needed ;)) |
| siggg (~jean@tra78-1-87-88-161-80.dsl.club-internet.fr) left irc: Ping timeout: 480 seconds |
| 16:26 Mrfai | no, yes '' | just move an infinite number of carriage-return into the pipe |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. |
| ErKa (keryell@m6.wifi.enstb.org) left irc: Ping timeout: 480 seconds |
| 16:42 FBI | fai: 3 lange committed revision 5407 to 7 fai: remove yes pipe when calling debootstrap |
| 16:47 mikap | Mrfai: ok |
| Action: mikap doesn't like the current changelog ;) |
| verwilst (~verwilst@d5152EF90.access.telenet.be) joined #fai. |
| 16:54 mgoetze | Mrfai: biste morgen um 10 im buero? oder doch besser nachmittags? :) |
| 16:55 Mrfai | besser nachmittags |
| 16:55 mgoetze | oki |
| kai_bo (~kai@statdsl-085-016-072-173.ewe-ip-backbone.de) left irc: Quit: Verlassend |
| Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving |
| ErKa (keryell@keryell.pck.nerim.net) joined #fai. |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Quit: Leaving. |
| verwilst (~verwilst@d5152EF90.access.telenet.be) left irc: Quit: Ex-Chat |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. |
| siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Quit: Leaving. |
| ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds |
| ErKa (keryell@keryell.pck.nerim.net) joined #fai. |
| verwilst (~verwilst@78-21-41-88.access.telenet.be) joined #fai. |
| ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds |
| 21:45 mikap | "Ok, aborting, since modules for this image already exist." [http://paste.pocoo.org/show/g4buK1n9fNqYVmhKRbC9/] => any ideas how to force FAI to install kernel updates? |
| 21:47 mikap | debconf preseeding? though how to automatically match the kernel version? |
| verwilst (~verwilst@78-21-41-88.access.telenet.be) left irc: Quit: Ex-Chat |
| --- Tue May 5 2009 |