[OE-core] postinst does not finish
Koen Kooi
koen at dominion.thruhere.net
Fri Apr 19 12:49:20 UTC 2013
Op 19 apr. 2013, om 14:26 heeft Laurentiu Palcu <laurentiu.palcu at intel.com> het volgende geschreven:
>
>
> On 04/19/2013 02:38 PM, Koen Kooi wrote:
>>
>> Op 19 apr. 2013, om 13:23 heeft Laurentiu Palcu <laurentiu.palcu at intel.com> het volgende geschreven:
>>
>>>
>>>
>>> On 04/19/2013 12:13 PM, Koen Kooi wrote:
>>>>
>>>> Op 19 apr. 2013, om 10:52 heeft "Burton, Ross" <ross.burton at intel.com> het volgende geschreven:
>>>>
>>>>> On 19 April 2013 09:38, Burton, Ross <ross.burton at intel.com> wrote:
>>>>>> On 18 April 2013 19:45, Koen Kooi <koen at dominion.thruhere.net> wrote:
>>>>>>> [koen at rrMBP 1.0-r0]$ grep "configuration required on target" temp/log.do_rootfs
>>>>>>> gdk-pixbuf-loader-png.postinst returned 1, marking as unpacked only, configuration required on target.
>>>>>> ..
>>>>>>
>>>>>> This is "usual" for postinsts that are intercepted and handled in a
>>>>>> single go - scroll further down the log and you should see the package
>>>>>> manager running the intercept scripts (i.e. fontcaches once, gconftool
>>>>>> once). I expect you'll see there that the gconftool intercept is
>>>>>> breaking for some reason.
>>>>>
>>>>> Actually gconf isn't using triggers but it should happen on the host.
>>>>>
>>>>> I wonder if we're hitting a problem where the postinsts are doing some
>>>>> actions on the host and then triggers are "erroring" out and it's all
>>>>> falling back to the target. Can you share the complete rootfs log for
>>>>> this gnome image?
>>>>
>>>> Here's the complete log: http://dominion.thruhere.net/koen/angstrom/beaglebone/log.do_rootfs
>>> Do you think you can post the rootfs/var/lib/opkg/status file? This
>>> would show us exactly what packages were postponed for first boot.
>>
>> This is the build that finished just before Ross sent the gconf patch:
>>
>> http://dominion.thruhere.net/koen/angstrom/beaglebone/Angstrom-systemd-GNOME-image-eglibc-ipk-v2013.06-beaglebone.rootfs.tar.xz (75MiB)
>>
>> I haven't booted that yet, I'll do that when I get back from lunch.
> From what you sent, I see the following packages postponed: udev-hwdb,
> gnome-panel and gdm. We know about udev-hwdb... However, gnome-panel and
> gdm are postponing the postinstalls for first boot explicitly:
>
> if [ "x$D" != "x" ]; then
> exit1
> fi
>
> This explains, at least, why you still see postinstalls running at first
> boot.
>
> Running the postinstalls at first boot appear to hang because (and
> that's just a feeling) gtk-icon-cache is run twice on target... which is
> time consuming.
I know it's time consuming, but it finished on 60-90 minutes with 1.3, I left it overnight and it still hadn't finished. I have the LEDS configured to short mmc0 and cpu0 activity and there' s no IO, only cpu from time to time. I don't think it takes more than 14 hours to run those postinsts.
regards,
Koen
More information about the Openembedded-core
mailing list