Started processes should be killed on SIGINT

[marianogappa]

No description provided.

[therealplato]

https://github.com/marianogappa/sql/pull/9/files

[marianogappa]

This is very good stuff! Tests show that even though the ssh processes are indeed killed, the queries on the remote server are not killed. Unfortunately, that detail makes the feature less than ideal. Let's spend a little more time to see if we can change ssh's behaviour on termination. Perhaps a different termination signal does the job.

[marianogappa]

I'll merge it and then start testing with ssh -t -t to see if the remote mysql processes are terminated.

[marianogappa]

It was more complicated than we thought. This is a very insightful response: https://serverfault.com/questions/463366/does-getting-disconnected-from-an-ssh-session-kill-your-programs/463375#463375?newreg=262ab662145f48e6b09d66fe21310cae

The mysql client process will terminate shortly. That doesn't mean at all that the mysql server will abort the query. We can look into a configuration in the mysql client exec call to make it abort queries on HUP, but there is nothing we can do from the ssh command.

[marianogappa]

Also I discovered today we can use the SLEEP(seconds) function on MariaDB to test this!

MariaDB [test]> select sleep(10);

10 seconds later…

+-----------+
| sleep(10) |
+-----------+
|         0 |
+-----------+

[marianogappa]

I'm afraid this is a limitation of the mysql client; I can validate the queries still live on the server until they finish. Research only points to using SHOW PROCESSLIST + KILL 1234. We can revise this in the future.