Bug#829180: appears to be failing because of fsck

[Daniel Pocock]

On 15/12/16 11:37, Michael Biebl wrote:
> Am 15.12.2016 um 07:31 schrieb Daniel Pocock:
>
>> Please look more closely ... 3 lines after that it says the fsck has
>> "Started" and then it gives another line about "cgroup is empty"
>>
>> Could that just be a fault with re-ordering the log entries in
>> journald?  Or maybe it is starting fsck twice?
>>
> No, that message is printed *after* the job has finished:
> Starting foo.service [job is about to be executed]
> ...do stuff...
> Started foo.service [job is finished]
>
> Note the Start*ing* vs Start*ed*
>
>


I had a closer look and used


    tune2fs -C 26 /dev/mapper/vg00-foo


to increment the mount count and force a check on the next boot


Watching the console as it booted, I notice the "Started...." line only
appears on the console as the fsck is finishing


The word "Started" seems out of place in this context


As suggested in the github discussion[1], I've put in a substitute
/bin/mount script to try and catch more details.  Can you suggest
anything else I should try either


a) to include in my /bin/mount script?

b) to test after the mount fails again?  (it doesn't happen every boot)


Regards,

Daniel



1. https://github.com/systemd/systemd/issues/4812#issuecomment-265928415