The SSH keepalive mechanism consists of two key parameters in client configuration:

ServerAliveInterval 60
ServerAliveCountMax 2

Contrary to some misconceptions, ServerAliveCountMax doesn't limit the number of keepalive packets sent during normal operation. Instead, it defines how many consecutive failed keepalive responses will trigger connection termination.

Here's the actual behavior:

• Every ServerAliveInterval (60s in example), client sends keepalive request
• If server responds, counter resets to 0
• If no response, counter increments by 1
• When counter reaches ServerAliveCountMax, connection drops

For different network conditions:

# Stable network (cloud servers)
Host cloud
    ServerAliveInterval 30
    ServerAliveCountMax 3

# Unstable network (mobile connections)
Host mobile
    ServerAliveInterval 15
    ServerAliveCountMax 10

To verify your configuration:

ssh -v user@hostname

Look for debug messages containing "Sending keepalive" to confirm the interval is working.

Remember that server-side may have its own timeout settings (ClientAliveInterval and ClientAliveCountMax in sshd_config). For persistent connections, both sides need proper configuration.

If connections still drop prematurely:

• Check for network devices (firewalls, NAT) that might drop idle connections
• Verify TCP keepalive settings at OS level
• Consider using persistent SSH connections with ControlMaster

# Example of persistent connection setup
Host *
    ControlMaster auto
    ControlPath ~/.ssh/control:%h:%p:%r
    ControlPersist 1h

The ServerAliveInterval and ServerAliveCountMax parameters work together to maintain SSH connections during periods of inactivity. Here's the technical breakdown:

Host *
    ServerAliveInterval 60   # Send keepalive every 60 seconds
    ServerAliveCountMax 2    # Max unanswered probes before disconnect

Contrary to some misconceptions, the behavior is precisely defined:

1. After ServerAliveInterval seconds of inactivity, the client sends an encrypted keepalive message
2. If no response is received, the client will wait another interval period before retrying
3. This continues until ServerAliveCountMax consecutive probes go unanswered
4. Only then does SSH terminate the connection

Consider these real-world cases with different configurations:

# Scenario 1: Aggressive keepalive (5 minute timeout)
Host production-server
    ServerAliveInterval 30
    ServerAliveCountMax 10  # 30×10 = 300s (5m) total timeout

# Scenario 2: Loose connection (1 hour timeout)
Host development
    ServerAliveInterval 300
    ServerAliveCountMax 12  # 300×12 = 3600s (1h) total timeout

• These are client-side settings (affects outgoing connections)
• For server-side keepalives, use ClientAliveInterval in sshd_config
• TCP keepalive (different mechanism) can be enabled with TCPKeepAlive yes

If keepalives aren't working as expected:

# Verify your settings with:
ssh -G yourhost | grep Alive

# For debugging, add verbose output:
ssh -vvv yourhost

For enterprise environments with strict requirements:

Match host *.corp.example.com
    ServerAliveInterval 15
    ServerAliveCountMax 4
    TCPKeepAlive yes
    ExitOnForwardFailure yes