tcpkeepaliveprobes

tcpkeepaliveprobes refers to the number of TCP keepalive probes that may be sent on an idle connection before the connection is considered dead. TCP keepalive is a mechanism used to verify that a peer is still reachable when no data has been exchanged for some time.

When a connection becomes idle beyond the configured keepalive idle time, the network stack may send a series of keepalive probes at regular intervals. If a probe elicits a response, the idle timer resets. If no response is received after the configured number of probes, the connection is closed as dead to free resources.

In Linux systems, the keepalive settings are exposed as sysctl values and per-socket options. The global defaults commonly include net.ipv4.tcp_keepalive_time, net.ipv4.tcp_keepalive_probes, and net.ipv4.tcp_keepalive_intvl (the time before starting probes, the maximum number of probes, and the interval between probes, respectively). Typical defaults are about 7200 seconds for keepalive_time, 9 probes, and 75 seconds between probes, though distributions may vary. These values can be viewed and modified with sysctl or by writing to /proc/sys/net/ipv4/, for example net.ipv4.tcp_keepalive_probes=5. At the per-socket level, applications can configure TCP_KEEPIDLE, TCP_KEEPINTVL, and TCP_KEEPCNT via setsockopt after enabling SO_KEEPALIVE.

Keepalive probes help detect stale connections in long-running or idle sessions but introduce additional network traffic and may be affected by intermediate devices such as NAT routers and firewalls. Some applications disable keepalives by default or tailor the values to balance timely detection with network overhead.