[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: systemd log messages during boot (Re: I'm not a huge fan of systemd)



On Wed, Jul 23, 2014 at 2:02 AM, Erwan David <erwan@rail.eu.org> wrote:
> On Wed, Jul 23, 2014 at 12:06:51AM CEST, Michael Biebl <biebl@debian.org> said:
>> Am 22.07.2014 19:22, schrieb Erwan David:
>>> Le 22/07/2014 18:59, Michael Biebl a écrit :
>>>> Am 22.07.2014 18:24, schrieb The Wanderer:
>>>>
>>>>> As far as I can see, there is no way to get init-system log messages
>>>>> without also getting kernel log messages
>>>> Of course there is.
>>>>
>>>> Might help if you actually tried it before commenting on it?
>>>>
>>>> The systemd.* specific flags override the global quiet flag. The
>>>>
>>>> So you can very well keep the quiet kernel command line argument and use
>>>>
>>>> systemd.show_status=true|false
>>>> systemd.sysv_console=true|false
>>>> systemd.log_level=...
>>>> systemd.log_target=...
>>>>
>>>> etc. to control in a very fine grained manner, how the data is logged.
>>>
>>> It would be interesting if the default was not changed, ie. same
>>> behaviour when using the default configuration.
>>
>> The default wasn't changed, really.
>> It's simply that SysV init scripts are so horribly inconsistent and
>> interpret the "quiet" parameter differently. So we don't have a
>> consistent behaviour wrt to logging and output.
>
> The defauklt was changed in that nomessage at all, no sign of any
> progression is NOT the former behaviour.
>>
>> The example skeleton SysV init script /etc/init.d/skeleton, which is
>> supposed to be a base for newly written init scripts uses
>> /lib/init/init-d-script. If you take a look at that script, you'll see
>> that prefixes its log message with [ "$VERBOSE" != no ] && log_*_msg
>>
>> And surprise, VERBOSE is set to "no" by /lib/init/vars.sh if the kernel
>> command line contains "quiet".
>>
>> Thankfully, this is all fixed now with systemd, where you have a
>> consistent and central place to configure that.
>
> NO it ids NOT fixed,k because what imports is NOT the theory but the
> actual behaviour. The actual behaviour is changed, and the new one is
> more than disturbing.

The behavior of the boot messages hasn't changed for me and according
to the systemd man pages it shouldn't. So your setup must be
different.


Reply to: