[13:57] hello. i got: init: mounted-proc main process (1277) terminated with status 1 [13:58] using ubuntu-core 12.04.1 on armv7 [14:03] where to find information about linux kernel features needed for upstart? [14:07] ericbutters: odd - that job just calls chmod. Is there anything in /var/log/upstart/mounted-proc.log* ? does /proc/eslabinfo exist? what perms does it have? [14:07] ericbutters: why do you need to know the kernel features? [14:08] i can't boot into ubuntu because of that error. [14:08] i built kernel by myself [14:08] i can boot into fedora 17 [14:10] jodh can not answer to your questions because i can not access the filesystem [14:23] my kernel command line: console=ttySAC1,38400 rootfstype=ext3 rw root=/dev/mmcblk0p2 rootdelay=5 --no-log noinitrd [14:24] ericbutters: you can remove --no-log now - upstart supports initramfs-less boot. [14:24] ericbutters: try booting with "init=/bin/sh" and having a look around your system maybe? [14:24] when i remove i get: https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/980917 [14:29] ericbutters: oh, sorry - hadn't seen you're running 12.04.1. yes, that SRU will be progressed soon :-) [14:29] jodh i now have /bin/sh stated... what should i look around? [14:34] well, remount root rw (http://upstart.ubuntu.com/cookbook/#boot-to-a-shell-directly) and then try mounting /proc maybe? You could then "exec /sbin/init" and see what you get. Why are you using your own kernel? Presumably, the system works normally with a stock Ubuntu kernel? If so, you should compare your kernel config with /boot/config-. [14:35] we are OEM, porting linux to our ARM based boards.. [14:37] FYI this is output: http://pastebin.com/raw.php?i=H4hTDAF0 [14:43] If the problem isn't with /proc, try adding --debug to your boot command-line. [14:47] jodh: when bootargs with --no-log --> http://pastebin.com/raw.php?i=V8pe9KDG and when bootargs without --no-log --> http://pastebin.com/icfUFgt0 [14:49] ericbutters: the first error is because /bin/sh exited as it doesn't understand --no-log - that's an upstart option. [14:50] can you boot with --debug? [14:50] first w/ debug and no --no-log: http://pastebin.com/yLs7G3wQ [14:51] secong with --no-log: http://pastebin.com/P9wc3g7h [15:00] ericbutters: you appear to have modified the jobs - 'mountallServer' is not a standard event. Also, have you compared your kernel config with the stock ubuntu ones? [15:01] jodh i did not anything to the ubuntu system. i compare with ac100 ubuntu defconfig at the moment. is not that easy. ac100 is 3.1.10 where my linux kernel is 3.3.7 [15:04] i can access by mounting the sd card on my devel machine. what to do to get this mountallServer disabled? [15:07] yo [15:08] ogra_: hi! [15:08] ericbutters: can you explain why you can't use the stock ubuntu kernel source + config and rebuild for your board maybe? [15:10] ericbutters: I've double-checked and nothing that I can see in an standard ubuntu 12.04.1 system emits the 'mountallServer' event, so your system has been customised somehow. Try booting with init=/bin/sh and doing grep -ilr mountallServer /etc/init* [15:10] for arm-specifics, you might also want to ask on #ubuntu-arm. [15:11] uuhh that takes too much time, getting all the samsung specific things to work.. all works great with fedora. i just want to try ubuntu. [15:13] ? I don't understand sort, grep and diff should allow you to compare those 3 kernel configs?? [15:14] so I would compare kernel configs, ensure you have inotify and /proc compiled into the kernel, and identify what introduced the 'mountallServer' event. [15:15] and devtmpfs [15:15] ogra_: yeah - and sysfs and tmpfs. [15:15] jodh grep -ilr mountallServer /etc/init* gives nothing [15:16] jodh ogra_ all that is enabled in kernel [15:17] well, do you aheva proper rootfs as well ? [15:17] *have a [15:17] ericbutters: gah - mountallServer is a "red-herring" it was added to mountall a while back and sadly doesn't conform to the standard naming scheme. [15:17] ericbutters: so, ignore that issue. [15:18] orga_ i have a fedora 17 lxde running [15:18] if all kernel optiuons are there, /me bets on an incomplete rootfs [15:18] ericbutters, i dont care about fedora, its so massively different from ubuntu, it doesnt help to mention it every few mins [15:18] i downloaded from ubuntu-core here: http://cdimage.ubuntu.com/ubuntu-core/releases/12.04/release/ [15:19] as i told you before [15:19] ubuntu-core is *NOT* a root filesystem to boot from [15:19] i pointed you to the right rootfs and also told you wheer to find a config in there to compare to your kernel [15:20] where can i find *A* ubuntu for arm to boot from? [15:20] http://cdimage.ubuntu.com/lubuntu/daily-preinstalled/20121008/quantal-preinstalled-desktop-armhf+ac100.tar.gz has a lubuntu tarball ... [15:20] orga_ thank you.. i will try.. did you read my first post regarding mounted-proc? [15:20] after untarring run: touch /var/lib/oem-config/run [15:20] ericbutters, most likely you are missing config options,. the tarball is originally for the ac100, look in /boot/ for the configuartion file and compare it to your config options [15:20] repatesd from #ubuntu-devel [15:20] *pasted [15:21] i wont repeat myself again [15:21] orga_ okay.. i try that one.. [15:21] sry ogra_ [15:22] ericbutters: another thing - if you haven't already created one, you'll probably want to create an upstart job that starts a getty on /dev/ttySAC1. [15:22] yeah [15:22] see the serial howto on the ubuntu wiki [15:22] okay [15:22] it describes the process to set serial console up [15:51] jodh ogra_ wow. i created ttySAC1.conf and added ttySAC1 to securetty.. and now i got a log in. sry, but what is the login name and passwd? [18:40] upstart 0.6.5 on el6: what's the recommended way to restart a process when any instance of another job has restarted (most liked due to a crash)? Can I specify a pattern in the "start on" and "stop on" clauses? [18:48] Hmm. maybe the instance job name without any INSTANCE= will work? [19:02] yes, that seems to work. [19:05] bfallik: glad we could help ;) [19:06] hehe. I'm still not 100% certain of the right way to do what I want, but at least I have my answer. [19:29] bfallik: note that upstart won't emit events, as far as I know, for a respawn of a service... so it would have to be the service dying, upstart not respawning (or giving up) and changing the goal back to stop.. and then an admin or other process starting it again. [19:59] SpamapS: Interesting. That doesn't match what I've observed. [20:00] we have respawn set in our worker job, and the webserver job has: start on starting worker; stop on (stopping worker or stopped worker) [20:01] it seems to DTRT, although I'm not expert enough to know why.