feat: redact secrets in debug logging #209
Conversation
pyrooka
force-pushed
the
nb/redact-secrets
branch
from
fad3f75 to
ed37a7c
Compare
| @@ -33,40 +30,7 @@ | |||
| key = os.path.join(os.path.dirname(__file__), '../resources/test_ssl.key') | |||
|
|
|||
|
|
|||
| def _local_server(tls_version: int, port: int) -> Callable: | |||
There was a problem hiding this comment.
I moved this to a separate file, so that it's easier to use it in other tests too.
padamstx
force-pushed
the
debug-logging
branch
4 times, most recently
from
e0a28a7 to
7b11e75
Compare
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
pyrooka
force-pushed
the
nb/redact-secrets
branch
from
ed37a7c to
a348672
Compare
There was a problem hiding this comment.
I'm testing out your feature branch with the platform-services python SDK and I've run into some problems related to the use of a redacted message. It looks like the log record contains both the fully-resolved/redacted message AND the individual arguments that were originally passed on the call to logger.debug() which are used to format the message. I think once we replace the log record's message with the fully-resolved/redacted version, then we should probably remove the arguments from the log record to avoid this error (presumably)
|
Update: After experimenting with your feature branch in my local sandbox, I opened a PR with a handful of changes (most of which we discussed on slack) that fix the issues that I ran into. |
Signed-off-by: Phil Adams <phil_adams@us.ibm.com>
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
Signed-off-by: Norbert Biczo <pyrooka@users.noreply.github.com>
This PR improves the debug logging logic in the core,
so that secrets are not shown in the debug logs anymore.
FYI: this PR targets the
debug-loggingfeature branch.