Make SIGHUP reload and update cache factor values

[erikjohnston]

This would be very useful when you want to update cache factors across many workers without having to restart them all (e.g. when they're under load and restarting would cause issues)

[DMRobertson]

Is this as simple as having a SIGHUP handler call this function?

synapse/synapse/config/cache.py

Lines 264 to 274 in 4586119

	def resize_all_caches(self) -> None:
	"""Ensure all cache sizes are up to date
	For each cache, run the mapped callback function with either
	a specific cache factor or the default, global one.
	"""
	# block other threads from modifying _CACHES while we iterate it.
	with _CACHES_LOCK:
	for cache_name, callback in _CACHES.items():
	new_factor = self.cache_factors.get(cache_name, self.global_factor)
	callback(new_factor)

[DMRobertson]

Oh no, we'd need to get the homeserver to reparse the config first.

[clokep]

We do something similar for logging config near

synapse/synapse/config/logger.py

Line 345 in 8d156ec

appbase.register_sighup(_reload_logging_config, log_config_path)

[DMRobertson]

That's a helpful pointer, thanks. WIll still have to work out how to wrangle our config class magic.

[clokep]

Yeah, although I realize now it isn't quite the same since the logging stuff has a separate config file.