log. _prefix_from_stack() needs a defined behaviour.
#185
Labels
bug
Something isn't working
Comments
terjekv
added a commit
to terjekv/mreg-cli
that referenced
this issue
Dec 6, 2023
## Changes - Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. ## Issues - We want timestamps in the logs, but they will differ for CI runs and thus fail. There needs to be a configuration option to disable logging timestamps... - The stack details for errors and warnings is bugged. See unioslo#185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ``` ## Failing tests As this completely redefines the recording output, tests fail miserably.
terjekv
added a commit
to terjekv/mreg-cli
that referenced
this issue
Dec 6, 2023
## Changes - Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. ## Issues - We want timestamps in the logs, but they will differ for CI runs and thus fail. There needs to be a configuration option to disable logging timestamps... - The stack details for errors and warnings is bugged. See unioslo#185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ``` ## Failing tests As this completely redefines the recording output, tests fail miserably.
terjekv
added a commit
to terjekv/mreg-cli
that referenced
this issue
Dec 11, 2023
## Changes - Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. ## Issues - We want timestamps in the logs, but they will differ for CI runs and thus fail. There needs to be a configuration option to disable logging timestamps... - The stack details for errors and warnings is bugged. See unioslo#185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ``` ## Failing tests As this completely redefines the recording output, tests fail miserably.
terjekv
added a commit
to terjekv/mreg-cli
that referenced
this issue
Dec 12, 2023
## Changes - Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. ## Issues - We want timestamps in the logs, but they will differ for CI runs and thus fail. There needs to be a configuration option to disable logging timestamps... - The stack details for errors and warnings is bugged. See unioslo#185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ``` ## Failing tests As this completely redefines the recording output, tests fail miserably.
terjekv
added a commit
to terjekv/mreg-cli
that referenced
this issue
Dec 13, 2023
- Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. - We want timestamps in the logs, but they will differ for CI runs and thus fail. There needs to be a configuration option to disable logging timestamps... - The stack details for errors and warnings is bugged. See unioslo#185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ``` As this completely redefines the recording output, tests fail miserably.
terjekv
added a commit
that referenced
this issue
Dec 18, 2023
* Reformat and redesign the recording output. - Formats command events into properly related JSON. - Properly record status responses (ok, warning, errors). - Record API requests as a chain related to each command. - Record the raw command issued, the actual command executed, and the command filter applied (if any). - Moves comment handling to OutputManager. - Consolidates stop_recording and save_recording. - Renamed some variables and methods for consistency. - Recording-related methods now all start with recording_ as a prefix. - Added a command line option to enable or disable timing data from the recorded data. - In general, callers adding details to the recorded data shouldn't test if recording is active. OutputManager handles that detail, the rest of the code simply states that it wishes to make something available for recording. * Issues - The stack details for errors and warnings is bugged. See #185 - We buffer the entire recording blob in memory until we stop recording (or we exit). This is to ensure valid JSON for the output, as writing (valid) JSON streams to a file is non-trivial. A solution is to write after each command, reload the file and append the data, then write again. This may or may not be a good idea, and it would look something like: ```python existing_data = [] # Read the existing data from the file try: with open(self._filename, "r") as rec_file: existing_data = json.load(rec_file) except (FileNotFoundError, json.JSONDecodeError): pass existing_data.append(recording_entry) # Write the updated data back to the file with open(self._filename, "w") as rec_file: json.dump(existing_data, rec_file, indent=2) ```
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
mreg-cli/mreg_cli/log.py
Line 13 in 4e3267b
The question is if this stack parsing should be exposed to the users or if it should be logged or available as debug information. Ideally we probably want messages back to the user to make sense for the user if it is something they can do differently. Right now, even warnings such as
host not foundtries to provide information from the stack and that seems counterproductive to say the least.The text was updated successfully, but these errors were encountered: