【Patroni源码阅读】处理启动过程中PG（启动超时或者需要failover）

handle_starting_instance.py

class Ha(object):
    def handle_starting_instance(self):
        """Starting up PostgreSQL may take a long time. In case we are the leader we may want to
        fail over to."""

        # Check if we are in startup, when paused defer to main loop for manual failovers.
        if not self.state_handler.check_for_startup() or self.is_paused():
            # pg 不是strartup状态或者patroni处于paused状态
            self.set_start_timeout(None)
            if self.is_paused():
                self.state_handler.set_state(self.state_handler.is_running() and 'running' or 'stopped')
            return None

        # state_handler.state == 'starting' here
        # 处理starting状态的pg
        if self.has_lock():
            # 当前节点是leader
            if not self.update_lock():
                # 更新Leader状态状态失败，这时可能该节点已经不是leader了
                # 此时立即停止pg
                logger.info("Lost lock while starting up. Demoting self.")
                self.demote('immediate-nolock')
                return 'stopped PostgreSQL while starting up because leader key was lost'

            # 依然是leader，根据超时时间判断启动逻辑
            timeout = self._start_timeout or self.patroni.config['master_start_timeout']
            time_left = timeout - self.state_handler.time_in_state()

            if time_left <= 0:
                # 超时，判断能不能进行主备切换，如果可以立即停止当前的Master
                if self.is_failover_possible(self.cluster.members):
                    logger.info("Demoting self because master startup is taking too long")
                    self.demote('immediate')
                    return 'stopped PostgreSQL because of startup timeout'
                else:
                    return 'master start has timed out, but continuing to wait because failover is not possible'
            else:
                # 还未超时，判断此时是否要failover发生，如果有正常退出postgres进程，并让出leader
                msg = self.process_manual_failover_from_leader()
                if msg is not None:
                    return msg

                return 'PostgreSQL is still starting up, {0:.0f} seconds until timeout'.format(time_left)
        else:
            # Use normal processing for standbys
            logger.info("Still starting up as a standby.")
            return None