On (03/12/14 11:01), Dmitri Pal wrote:
On 12/03/2014 10:56 AM, Pavel Březina wrote:
>On 12/02/2014 10:32 PM, Jakub Hrozek wrote:
>>On Tue, Dec 02, 2014 at 04:20:17PM -0500, Dmitri Pal wrote:
>>>On 12/02/2014 04:14 PM, Jakub Hrozek wrote:
>>>>On Tue, Dec 02, 2014 at 04:00:33PM -0500, Dmitri Pal wrote:
>>>>>Do we have any place where we describe what level of output one
>>>>>with each level?
>>>>sssd.conf has some info:
>>>This does not show how bitmap values map levels.
>>>IMO from usability POV levels a simpler though bitmaps are more
>>>If we want people to stop using 1-10 levels we need to stop
>>>them and start recommending bitmaps.
>>We tried, but bitmaps are just too hard to use for most users. So we
>>added this sentence to the man page:
>That is one thing, the other thing is that we are still quite inconsistent
>with the bit maps in the old code where numeric levels were used. The
>numbers were converted only on syntax level not the semantic one.
>One day, we may want to make the bitmask more refined. For example we can
>move too low level messages (such as sbus_toggle_watch, ldb tevent, ...) to
>a separate level because those are not usually needed but other "level 9"
>messages can still be helpful.
Let us focus on the user in this case.
I am a user.
Please give me the guidance what debug level I should use.
I general high level instruction will be good enough.
I want to troubleshoot a problem, what level should it use?
9? 10? Bit mask? If I do 7 is it enough?
Should I just always use the highest so that you get everything once?
Going several rounds is probably the most annoying thing.
I would recommend to use
debug_level = 7.
It is not simple to read log files therefore I would also recommend to filter
the most critical errors from obtained log files.
grep -E "(0x00[1-9])" sssd_sth.log
If it does not help then you need to find wider context around problematic
debug messages obtained from grep.