Make the external logging queue discard severity configurable

[quanah]

SUMMARY

Assisted-by: Claude Opus 5.5. Drafted with AI assistance; the commit carries the same trailer.

construct_rsyslog_conf_template() emits queue.discardSeverity="5" as a literal. Job events and activity stream entries are logged at INFO, so once LOG_AGGREGATOR_LEVEL lets them through and the action queue reaches queue.discardMark, all of them can be dropped, silently, and no setting can change that.

Register LOG_AGGREGATOR_ACTION_QUEUE_DISCARD_SEVERITY (0–8, default 5) and emit it in place of the literal. The default is what Ascender emits today, so existing configurations render byte-identically and test_what_is_discarded_is_the_least_important still holds; 8 disables discarding, at the cost of blocking the sender for queue.timeoutEnqueue when the queue is genuinely full. The setting is also added to the durability test's list of user-visible settings that must reach the queue.

Same change as ansible/awx#16714, ported to Ascender's generator.

related #1095

ISSUE TYPE

• Bug, Docs Fix or other nominal change

COMPONENT NAME

• API

ASCENDER VERSION

awx: 25.6.2-235-g640cf5f2

ADDITIONAL INFORMATION

Tests: test_rsyslog_conf_discard_severity in ascender/main/tests/unit/api/test_logger.py, parametrized over the registered default (which must still render 5), 8 and 0; plus one entry in test_the_settings_users_can_set_reach_the_queue (test_external_logging_durability.py), which records the setting as a promise and passes on main too. Run in python:3.14 with Ascender's requirements, as an unprivileged user:

# with this change
$ pytest ascender/main/tests/unit/api/test_logger.py ascender/main/tests/unit/utils/test_external_logging_durability.py
34 passed

# main with only the test changes
2 failed, 32 passed    (the 8 and 0 cases; the default case passes on main, which is the point of including it)

# generated action, LOG_AGGREGATOR_PROTOCOL=tcp, default settings: identical before and after
action(type="omfwd" ... queue.discardMark="117964" queue.discardSeverity="5")

# after, LOG_AGGREGATOR_ACTION_QUEUE_DISCARD_SEVERITY=8
action(type="omfwd" ... queue.discardMark="117964" queue.discardSeverity="8")

# rsyslogd -N1 on rockylinux:9, rsyslog-8.2510.0-2.el9_8.2, with discardSeverity="8": rc=0

ruff check and ruff format --check (ruff 0.16.6, as pinned in tox.ini) are clean on the touched files.