Description

When Splunk(HEC) is unreachable and ALLOW_WITHOUT_STRICT_LOGGING: true is set, the error logging in splunk_logs_model.py line 197 loses all context (kind, account, performer, repository) from the log message, making it impossible to know which action failed.

Version-Release number

Quay 3.17

How reproducible

Always

Steps to Reproduce

1. Configure Quay with LOGS_MODEL: splunk, producer: splunk_hec, and ALLOW_WITHOUT_STRICT_LOGGING: true
2. Make the Splunk(HEC) endpoint unreachable (e.g. stop Splunk or block port 8088)
3. Perform any action that generates an audit log (login, create org, push image, etc.)
4. Check the Quay app pod logs

Actual results

The pod log shows:

gunicorn-web stdout | [ERROR] [data.logs_model.splunk_logs_model] log_action failed

No context is included - you cannot tell which action failed, which user triggered it, or which namespace/repository was involved.

The root cause is at data/logs_model/splunk_logs_model.py line 197:

logger.exception("log_action failed", extra=({"exception": lse}).update(log_data))

dict.update() modifies the dict in-place and returns None in Python. So extra=None is passed to logger.exception(), and all the context fields (kind, account, performer, repository, ip) are lost.

Expected results

The error log should include the full context of the failed action, e.g.:

[ERROR] log_action failed | kind=create_repo account=testorg performer=quay repository=testrepo

Fix

extra = {"exception": lse}
extra.update(log_data)
logger.exception("log_action failed", extra=extra)

Additional info

Introduced by PR #4941 (PROJQUAY-10317).

Verified on a live Quay 3.17 instance with Splunk HEC at 18.116.40.103:8088 down. Both quay-quay-app pods show the bare log_action failed message with no useful details.