Fix wazuh-metrics for reconnecting daemons using multiprocessing

[vicferpoy]

In #2258, we adapted the wazuh-metrics CLI to allow monitoring child processes as multiprocessing was implemented for the Wazuh API and cluster.

However, with the current design it was not possible to keep monitoring said processes if they restarted:

wazuh-qa/deps/wazuh_testing/wazuh_testing/tools/performance/binary.py

Lines 177 to 190 in eb32b09

	except psutil.NoSuchProcess:
	logger.warning(f'Lost PID for {self.process_name}. Trying to obtain a new one. '
	'If the process has child processes, this test will not be valid')
	try:
	# Try to get another PID for the current process name. This could be wrong if there is more than
	# one process with the same name (child processes)
	self.pid = Monitor.get_process_pids(self.process_name, check_children=False)[0]
	self.set_process()
	except ValueError:
	logger.warning(f'Could not obtain a new PID for {self.process_name}. Trying again in {self.time_step}s')
	finally:
	info.update({key: round(value, 2) for key, value in info.items() if isinstance(value, (int, float))})
	logger.debug(f'Recollected data for process {self.pid}')
	return info

This is needed now as it is not compatible with our API performance tests due to their restart endpoints:

• api.log

2021/12/23 18:29:01 INFO: unknown_user 172.31.39.92 "GET /security/users/me/policies" with parameters {} and body {} done in 0.013s: 400
2021/12/23 18:29:24 INFO: Listening on 0.0.0.0:55000..
2021/12/23 18:29:31 INFO: wazuh 172.31.32.132 "GET /" with parameters {} and body {} done in 4.788s: 200
2021/12/23 18:29:32 INFO: wazuh-wui 172.31.39.92 "GET /manager/stats/analysisd" with parameters {"pretty": "True"} and body {} done in 0.135s: 200
2021/12/23 18:29:32 INFO: wazuh 172.31.32.132 "PUT /cluster/restart" with parameters {} and body {} done in 0.165s: 200
2021/12/23 18:29:35 INFO: unknown_user 172.31.39.92 "GET /cluster/status" with parameters {} and body {} done in 0.005s: 400
2021/12/23 18:29:58 INFO: Listening on 0.0.0.0:55000..

• One of the API processes CPU usage

wazuh-apid_child_1,4.3.0,2021/12/23 18:28:53,20389,40.1,342528.0,90220.0,63012.0,70317.0,0.0,28,4455,208,1621.0,152.0,12.8,0.0
wazuh-apid_child_1,4.3.0,2021/12/23 18:28:58,20389,21.5,345484.0,93052.0,65876.0,73163.0,0.0,28,6061,308,1625.0,352.0,0.8,40.0
wazuh-apid_child_1,4.3.0,2021/12/23 18:29:03,20389,-721.1,242944.0,33060.0,28516.0,28881.0,0.0,16,833,15,0.0,0.0,-325.0,-70.4
wazuh-apid_child_1,4.3.0,2021/12/23 18:29:08,20389,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0
wazuh-apid_child_1,4.3.0,2021/12/23 18:29:13,20389,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0

A rework of both the CLI and the Monitor class is needed to solve this.