| kg | just tried to boot 10.10 beta1 on e-machines E443 (amd E350 dual core) stopped blank screen after loading bluetooth something??? | 08:46 |
|---|---|---|
| kg | ops that should be 11.10 | 08:47 |
| kg | any ideas on how to proceed? | 08:49 |
| === McPeter_ is now known as McPeter | ||
| === manjo` is now known as manjo | ||
| jamespage | is it possible to see previous ISO testing results through iso.qa.ubuntu.com? I need to check when the iSCSI root tests last passed.... | 14:42 |
| hggdh | no, it is not, as far as I know. But IIRC, at least at alpha1 it was OK | 14:43 |
| hggdh | and I actually think all the alphas made it through with iSCSI | 14:44 |
| hggdh | jamespage: ^ | 14:44 |
| jamespage | hggdh - yeah thats what I thought - I think I ran them in the last two | 14:45 |
| jamespage | hence why I picked up beta | 14:45 |
| jamespage | stgraber: ^^ re last successful test of iSCSI root in Oneiric | 14:45 |
| hggdh | I think manual sql magic can do it, but I do not have access to the system | 14:46 |
| stgraber | jamespage: ok | 14:46 |
| jamespage | hggdh; crazy..... | 14:46 |
| stgraber | jamespage: my last test was with Natty's kernel and natty's open-iscsi on Oneiric and I still had the bug | 14:46 |
| jamespage | stgraber: yes - I just saw your update | 14:47 |
| stgraber | jamespage: I'm now trying with an up to date Natty just to make sure I can get a working setup at least once... | 14:47 |
| stgraber | ok, at least I get the exact same behaviour on natty, that's a start | 14:53 |
| stgraber | well, my current test isn't exactly the one you did, I'm actually doing: | 14:54 |
| stgraber | - regular ubuntu server install (on local disk) | 14:54 |
| stgraber | - install open-iscsi | 14:54 |
| stgraber | - scp config from my iscsi install on Oneiric into /etc/iscsi | 14:54 |
| stgraber | - update-initramfs (so that the initramfs initializes the iscsi link at boot time) | 14:54 |
| stgraber | - move /etc/init.d/open-iscsi aside | 14:54 |
| stgraber | - reboot | 14:54 |
| stgraber | I correctly see the connection from the iniramfs and it succeeds | 14:55 |
| stgraber | once booted, I start open-iscsi and notice that it breaks the established session and never reconnects | 14:55 |
| stgraber | well, it tries to reconnect but fails because of AuthMethod | 14:55 |
| stgraber | on a regular iscsi install, that'd appear as the system hanging half way through boot (at the moment where open-iscsi is started) | 14:56 |
| stgraber | I'm going to do a regular iscsi install with natty just to confirm | 14:56 |
| stgraber | jamespage: I'm actually starting to wonder if the problem isn't server side instead. Do you know if the last succesful test was also done using lucid + the config from the wiki or if it was using something more recent? | 15:02 |
| jamespage | stgraber: I think I did alpha3 against a lucid target | 15:02 |
| stgraber | ok, I guess I'll start downloading alpha3 then, once I'm done running my test on natty | 15:03 |
| jamespage | stgraber: I tried disabling the open-iscsi daemon on the images I created during beta1 testing; the unauthenticated one now boots - and quickly | 15:07 |
| jamespage | just testing the other now | 15:07 |
| jamespage | (as you described in the bug report really) | 15:08 |
| jamespage | ditto on the authenticated | 15:14 |
| stgraber | hmm, ok, really weird. Wondering if it's not just a case where open-iscsi won't start on older installs :) | 15:18 |
| jamespage | stgraber: well maybe | 15:19 |
| jamespage | stgraber: I wonder if the iscsi-network-interface upstart job is not doing its job right | 15:28 |
| jamespage | stgraber: or not firing early enough | 15:28 |
| jamespage | stgraber: still around - I have made a little progress in tracking down what the root cause might be | 16:11 |
| jamespage | I think this upstart change is the cause of the problem: http://launchpadlibrarian.net/76962956/upstart_1.3-0ubuntu5_1.3-0ubuntu6.diff.gz | 16:13 |
| jamespage | reverting to the previous configuration for rc-sysinit.conf - start on filesystem and net-device-up IFACE=lo | 16:14 |
| jamespage | causes the open-iscsi init script to fire earlier? maybe - and it runs through with IFACE lo which causes it not to start | 16:22 |
| stgraber | jamespage: oh, that'd explain it ;) so the bug is that it started working as designed? :) | 16:26 |
| jamespage | stgraber: well maybe - I'm still scratching my head a little | 16:26 |
| jamespage | but if I revert that upstart change to rc-sysinit.conf - I get a running system with open-iscsi | 16:27 |
| jamespage | sorry - without | 16:27 |
| * jamespage rubs his eyes | 16:27 | |
| jamespage | stgraber: what I can't figure out is why its running with something other than IFACE = lo when we see the issue | 16:28 |
| === yofel_ is now known as yofel | ||
| jamespage | stgraber: so with that upstart change in place open-iscsi also runs for eth0 | 16:58 |
| jamespage | it managed to boot but lots of io errors on the root disk | 16:59 |
| stgraber | jamespage: trying on natty confirms that back then iscsid just wasn't started at all | 17:10 |
| jamespage | stgraber: yep - thats what I see when I backout that upstart change - I updated the bug report. | 17:10 |
| jamespage | with details | 17:10 |
| stgraber | not exactly sure what's the fix for this though. I definitely don't have the required knowledge to fix iscsid to do the right thing and reverting that upstart change would be wrong too. | 17:12 |
| stgraber | cjwatson, SpamapS: any opinion ^ (bug 838809) | 17:12 |
| stgraber | no bot? https://bugs.launchpad.net/ubuntu/oneiric/+bug/838809 | 17:12 |
| stgraber | jamespage: a workaround would be to run "iscsiadm -m session" early in open-iscsi's init script and "exit 0" if it returns something. But that still feels wrong as I'd expect iscsid to start and just keep the existing sessions... | 17:16 |
| cjwatson | with iscsi root, iscsi is meant to be running from the initramfs. the main job of the init scripts is to detect that and leave well alone | 17:16 |
| stgraber | cjwatson: is iscsid supposed to be running? because I never saw it running post-initramfs. The iscsi session is established and working (maintained by the kernel module) but the management daemon isn't running. | 17:17 |
| stgraber | using the workaround I described before would work for cases where you have an iscsi root and no other mounted lun, if a user uses iscsi root and has additional iscsi luns defined in /etc/iscsi these won't be connected as iscsid won't be started at boot time. | 17:18 |
| stgraber | if that's not something we support, then I'm more than happy to do the required two lines change to the init script and be done with that bug ;) | 17:19 |
| cjwatson | iscsid> erm. I don't *think* so but I'm not certain | 17:19 |
| cjwatson | iscsistart is a little twisty | 17:19 |
| cjwatson | I think we need to support that | 17:19 |
| cjwatson | does iscsid itself fiddle with the session that / depends on, or is it the other stuff in the init script that does that? | 17:21 |
| stgraber | iscsid seems to try to logout from all sessions and login again on startup | 17:27 |
| stgraber | I can reproduce that starting it with: /sbin/iscsid -d 3 -f | 17:27 |
| stgraber | so it's the daemon itself doing it, not the init script | 17:28 |
| stgraber | I'm not entirely sure what's generating the iscsi configuration to start with. I'd think the initramfs only needs /etc/iscsi/iscsi.initramfs and we shouldn't have any other iscsi configuration on the target system | 17:29 |
| stgraber | so maybe the installer is generating some configuration files it shouldn't generate, causing open-iscsi to connect to something that's already connected instead of just leaving it alone | 17:31 |
| cjwatson | it's possible we're copying in duplicate configuration to the root fs that didn't matter before | 17:33 |
| cjwatson | you could check /target/etc/iscsi/ just before reboot into the new system | 17:34 |
| stgraber | I'm trying to start iscsid with an empty /etc/iscsi now to see what it does, hopefully it should start and sit there doing nothing | 17:34 |
| stgraber | argh, it doesn't... it tries to deal with the existing connection even though it's not anywhere in its config | 17:35 |
| stgraber | cjwatson: http://www.stgraber.org/download/iscsi-bug.png | 17:36 |
| stgraber | this is on my natty system when manually starting open-iscsi but I have the same happening on oneiric (should probably switch to it again now that we now where the problem is) | 17:37 |
| czajkowski | aloha | 18:00 |
| GrueMaster | Does anyone have a test/testsuite for SE Linux that I could run on ARM (i.e. no KVM/libvirt)? | 18:08 |
| davmor2 | czajkowski: hello | 18:17 |
| gbuonfiglio | Hello! I'm testing 11.10 but I'm experiencing a problem after running "unity --reset" and need some guidance. | 18:33 |
| gbuonfiglio | unity is missing icons, the panel looks weird (a bit like old gnome panel) and the last icon on the right side of the panel is inaccessible so I cannot logout or change system-settings for example. | 18:36 |
| gbuonfiglio | How can I reset everything from compiz to unity? | 18:37 |
| gbuonfiglio | I've created another user and everything looks fine from this user, so it looks like something is corrupted in my current user compiz or unit settings. | 18:43 |
| gbuonfiglio | I've just realized that other settings are messed up too. Keyboard was US-International and now is US. Reset gconf? | 18:47 |
| gbuonfiglio | unity-panel-service(5417): Operation not permitted | 18:55 |
| gbuonfiglio | I'm trying to run "gconftool-2 --recursive-unset /apps/compiz-1 && unity --reset" | 18:56 |
| gbuonfiglio | then I get several errors like this: unity.launcher LauncherIcon.cpp:408 Unable to load 'user-home' from icon theme | 18:56 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
