Huh. So IIUC, there's just more traffic than the site can currently handle? And that's why we have to essentially "wait our turn" to get a socket connection, by refreshing several times?
Still nothing but a
theory, mind. To put it into perspective, I'm standing behind Lalli right now, hear him shout "ONNI!" and an eerie voice answering "GO AWAY!", and guesstimate that Onni might be hard at work and has a bunch of (real world) customers waiting in line to get served.
If I looked up SYN, RST, and so forth via Google or glossary, would I get a usable definition?
Dunno ... In order to "open a (TCP) connection", as your browser tries to do, there needs to be a "three-way handshake" done. The client side sends a packet with no data and (only) the SYN flag set, the server replies with an (empty) packet with SYN and ACK(nowledge), and the client should reply in a timely fashion with an ACK (still empty if there's no data to be sent yet). From there on, packets get sent back and forth with ACK (and possibly PSH (push) for "urgent" data) set. When the time comes to terminate the connection, another three-way handshake is done, only with the intent being signaled with a FIN flag instead of SYNs.
RSTs (resets) terminate a TCP connection in an
irregular manner and
without a full three-way handshake, that's why all sorts of hard- and software that are not really one of the communicating parties - like firewalls or load-limiting mechanisms - throw those at the communicating parties (with "forged" sender and recipient coordinates, if need be) to cause them to drop (or stop initiating) the connection.
