不在同一服务器上的数据库连接起来久了会断开，大佬们有遇到过吗？

[liangmishi]

服务器 A 是 tornado，去连服务器 B 上的 PostgreSQL 数据库，刚开始没啥问题。
可是时间久了服务器 A 和服务器 B 的数据库连接就断了。 大佬们有遇到过这种情况吗？

附个错误提示：

Traceback (most recent call last):
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3756, in execute_sql
    cursor = self.get_cursor()
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3704, in get_cursor
    return self.get_conn().cursor()
psycopg2.InterfaceError: connection already closed

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/xxx/.local/lib/python3.5/site-packages/tornado/web.py", line 1488, in _execute
    result = self.prepare()
  File "/home/xxx/cc_backend/view/__init__.py", line 217, in prepare
    if not self.current_user():
  File "/home/xxx/cc_backend/view/__init__.py", line 192, in current_user
    return User.get_by_key(key)
  File "/home/xxx/cc_backend/model/user.py", line 102, in get_by_key
    return cls.get(cls.key == key)
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 4916, in get
    return sq.get()
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3171, in get
    return next(clone.execute())
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3223, in execute
    self._qr = ResultWrapper(model_class, self._execute(), query_meta)
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 2902, in _execute
    return self.database.execute_sql(sql, params, self.require_commit)
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3765, in execute_sql
    self.commit()
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3588, in __exit__
    reraise(new_type, new_type(*exc_args), traceback)
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 135, in reraise
    raise value.with_traceback(tb)
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3756, in execute_sql
    cursor = self.get_cursor()
  File "/home/xxx/.local/lib/python3.5/site-packages/peewee.py", line 3704, in get_cursor
    return self.get_conn().cursor()
peewee.InterfaceError: connection already closed

最后解决办法就是：每次使用完后就显式地关闭数据库。
关闭后，连接就会放回进程池，等到要用的时候 peewee 会自动去取。

[firefox12]

本来就是这样，所以有 mysql_ping 这样的设计

[SlipStupig]

太正常了，数据库会把一些闲置的连接主动杀掉（多长时间没有活动），可以通过 Keepalive 来连接，linux 可以通过一下参数来进行配置：
net.ipv4.tcp_keepalive_time=90
net.ipv4.tcp_keepalive_intvl=15
net.ipv4.tcp_keepalive_probes=2

但是依然会断开,postgresql 可以定时发送 NOTIFY 来保持连接

[msg7086]

数据库本身也有闲置断线设定，没记错的话 MySQL 是 3 小时，需要改配置延长时间。

[woshixiaohao1982]

难倒不使用线程池来维护连接对象？

[liangmishi]

@firefox12 原来如此，谢谢

@SlipStupig 增加连接时常的配置有点治标不治本的意思，我先谷歌看看 NOTIFY，谢谢

@msg7086 想说有没有一种断了再重连的方法，或者使用前尝试重连就可以解决了

@woshixiaohao1982 好主意，我去试试，谢谢

[QQ2171775959]

这个要看你怎么搭建了，服务器之间有没有组内网，设置的闲置时间有多长，有没有设置定时发送一条指令来激活这个闲置的连接。

[lostsquirrelX]

https://raw.githubusercontent.com/lostsquirrel/futures-collector/develop/simpletor/torndb.py
126 行

[liangmishi]

@QQ2171775959 感谢，学习了

@lostsquirrelX 感谢，学习了