Coding Rules Remote Cleanup

Remote Process Cleanup

Core Principle

When terminating local processes (SSH client, database client, etc.), must simultaneously terminate source processes on remote server triggered by that connection. Only terminating local connection will cause server resource exhaustion or even crash.

Trigger Scenarios

1. SSH Connection Interruption

• Before kill local SSH process, must first terminate remote executing commands through that connection or new connection
• If SSH executed database query through docker exec, need to first KILL at database level

2. Database Query Termination

• Before canceling local database client, must first execute KILL <process_id> on remote
• For queries executed through SSH tunnel, need to login database through new connection to terminate residual queries

3. Background Task Cancellation

• When canceling remote background task initiated locally, must confirm remote process terminated
• Cannot rely solely on local kill to terminate remote process

Standard Cleanup Process

1. Identify Remote Process: Through SHOW PROCESSLIST or ps aux confirm remote process ID
2. Terminate Remote Process: Execute KILL command through new connection
3. Verify Cleanup Result: Query again to confirm remote process disappeared
4. Terminate Local Process: After confirming remote cleaned, then terminate local connection

MySQL-Specific Cleanup

-- 1. View current active queries
SHOW PROCESSLIST;
-- Or more precise:
SELECT id, time, state, LEFT(info, 100) 
FROM information_schema.processlist
WHERE command != 'Sleep' AND time > 30;

-- 2. Terminate slow query
KILL <process_id>;

-- 3. Verify
SHOW PROCESSLIST;

Forbidden Behaviors

• ❌ Forbidden to only kill local process without cleaning remote
• ❌ Forbidden to assume "remote will auto-terminate after disconnect"
• ❌ Forbidden to initiate new query without confirming remote cleanup complete

Preventive Measures

• Set timeout for large table queries: SET SESSION max_execution_time = 60000;
• Avoid executing full table scans or no-index JOINs on remote
• For potentially time-consuming queries, first use EXPLAIN to evaluate
• Prioritize using LIMIT + sampling instead of full scan