site_archiver@lists.apple.com Delivered-To: darwin-kernel@lists.apple.com Hello all, Thanks, Matt Darland _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.a... Scenario: 1. Via the KPI interfaces a TCP connection established to a remote machine. 2. Normal read/write flow of data occurs. 3. Peer goes away without performing TCP teardown (e.g. goes to sleep). At this point, sock_sendmbuf() will happily take additional data, without returning an error. As of waiting about 3 minutes, none of the socket filter events get called. Is the above behavior known to occur (i.e. should I be checking something else in my code)? Does the TCP stack believe the connection is still valid? Is there some way at the KPI level I can detect when TCP believes the connection to be dead, or am I not waiting long enough? This email sent to site_archiver@lists.apple.com