Am 29.11.2016 um 12:22 schrieb Florent B:

On 11/29/2016 12:03 PM, Reindl Harald wrote:

...

Am 29.11.2016 um 11:49 schrieb Florent B:

...

For example, if I try this in PHP:

1/ Connect to mariadb 2/ sleep 610 seconds 3/ ping mariadb => I got Server has gone away error. With a 590 sleep time, no error. I use MariaDB 10.1, didn't change any "timeout" value in MariaDB : interactive_timeout & wait_timeout are 28800

http://php.net/manual/en/mysqlnd.config.php#ini.mysqlnd.net-read-timeout

Hi, and thank you. But I cannot find any timeout of 600 seconds... default_socket_timeout = 60 mysqlnd.net_read_timeout = 31536000

And I have the exactly same problem using this algorithm in Python, it's not related to PHP...

then pretty sure you mysqld timeouts are not what you think, otherwise we would see the same issues on our dbmail machines when after hours of idle postfix tries to re-use it's connections (and we did have such events when that values where too low) so why don't you ask the server instead pretend they are not lowered? MariaDB [(none)]> show variables like 'wait_timeout'; +---------------+-------+ | Variable_name | Value | +---------------+-------+ | wait_timeout | 300 | +---------------+-------+ 1 row in set (0.00 sec) MariaDB [(none)]> show variables like 'wait_timeout'; +---------------+--------+ | Variable_name | Value | +---------------+--------+ | wait_timeout | 604800 | +---------------+--------+ 1 row in set (0.00 sec) wait_timeout = 604800 interactive_timeout = 604800