I was looking at an issue today and noticed that the server never
really flips an SBH into SERVER_BUSY_READ after the first request
under worker.
ap_process_http_sync_connection:
ap_update_child_status_from_conn(c->sbh, SERVER_BUSY_READ, c);
while ((r = ap_read_request(c)) != NULL) {
...
ap_update_child_status(c->sbh, SERVER_BUSY_KEEPALIVE, NULL);
...
}
I was thinking we could at least transition to this state when
read_request_line() returns -- although non-malicious delays after
that seem pretty unlikely it might make for a better mod_status when
something unusual is happening.
(event changes to SERVER_BUSY_READ when when the request line is readable)
--
Eric Covener
[email protected]
