If a push fails, prometheus may never be reload even though config changed (theoretical edge-case)

[sed-i]

If some files were pushed but others failed, when _configure re-enters then should_reload may be false without prometheus ever reloading.

prometheus-k8s-operator/src/charm.py

Lines 292 to 304 in b4c0fd6

	try:
	# Need to reload if config or alerts changed.
	# (Both functions need to run so cannot use the short-circuiting `or`.)
	should_reload = any(
	[
	self._generate_prometheus_config(self.container),
	self._set_alerts(self.container),
	]
	)
	except PebbleError as e:
	logger.error("Failed to push updated config/alert files: %s", e)
	self.unit.status = early_return_statuses["push_fail"]
	return

[PietroPasotti]

we should probably let the error bubble up in this case, assuming the caller checked container.can_connect before this happens. Error state seems appropriate because this really shouldn't happen.

[simskij]

Issue has been stale for more than a year, and it's relevance today is thus debatable. Closing for now. Create a new issue or reopen this one if this is affecting you.