Hello.!
The current default appender is right for all our logs, except for shell
commands that are used by our dev-ops tools and generate very long (and
very useless) thread names".
How can we filter these commands that have no package, thus no explicit
appender ?
Another very strange shell behavior :
executing the command :
x="json = {\"a\":1, \"b\":2}"
.
produces a repeated "json = " pattern, it seems that the "{...,..." is
interpreted as a repeated action pattern. I've found no explanation in the
shell manual.
json = "a":1 json = "b":2
Is there a way to disable this command expansion : our dev-ops jobs fail
because of that :(
Thanks.
Regards.
Le mer. 1 oct. 2025 à 15:44, Matt Pavlovich <[email protected]> a écrit :
> Your logging configuration must be using a pattern that includes the
> thread name in the log output. You can remove that macro, or configure a
> separate log appender for the packages you want to filter and give that a
> different logging pattern without the thread macro.
>
> -Matt
>
> > On Oct 1, 2025, at 5:35 AM, Ephemeris Lappis <[email protected]>
> wrote:
> >
> > Hello.
> >
> > We need to use the Karaf's shell "log" command to trace actions during
> some deployment opérations..
> >
> > We've seen that the thread name using the "log:log" command is the
> command itself, producing very big lines in the log file.
> >
> > Example :
> > admin@root()> log:log --level WARNING "A very very long text..."
> > 12:29:39.362 WARN [pipe-log:log --level WARNING "A very very long
> text..."] A very very long text...
> >
> > The thread name is "pipe-log:log --level WARNING "A very very long
> text...""
> >
> > In reality, messages may be actually bigger, since we want to trace very
> detailed information about the current deployments.
> >
> > Is there any way to avoid this ?
> >
> > Thanks in advance for your help.
> >
> > Regards.
>
>
