Open beraldoleal opened 2 years ago
The most viable option so far seems to be to just move this to /var/log/avocado
(or ~/avocado/log
is installed locally) skipping the syslog altogether since many of could quickly flood it with logging messages that might not be appropriate for system level logging and it could be harder to filter these out from each other for later debugging. In comparison, we could always do a later consultation with a known avocado-generic log file similarly to the avocado-generic config file we are all aware of.
Since we are in Feature Freeze this week, moving this to the next sprint.
Is your feature request related to a problem? Please describe.
So far, Avocado was job-centric, where log messages are placed into
~/avocado/job-results/
per job. Those files are created at job time (when the Job() object is configured), and things like "self.job_log" are properly set.Due to the new Avocado distributed and multi-components architecture, it would be nice to log things even before the job has started.
Describe the solution you'd like Now that we are using the 'avocado.*' namespace for our logs, one possible suggestion would be to use
logging.handlers.SysLogHandler()
as an additional handler. Regardless of our current configuration (where the streams are saved) a system-wide logger service (syslog/rsyslog) would be able to catch those streams.Describe alternatives you've considered Another alternative is not using the SysLogHandler (or keep using it), but we pre-define a specific location and we need to care about the file creation outside of the job scope (i.e
~/avocado/log
).One solution doesn't invalidate the other.
Additional information See #5120, #4856