I’ve been getting lots of 408 timeouts when trying to access the SDMB recently (last day or two) - is it just me, or is anyone else getting this?
I get “Server Error” sometimes (one today), “Database Error” sometimes, but they are short-lived and rarely repeat immediately. I’ve sent TubaDiva screen shots, and she forwards them to Jerry, but nothing ever gets reported back. So the official word is shit happens.
I get occasional database errors but not often, maybe a couple per week.
Try doing a tracert from a command prompt (assuming you are on a Windows computer). That may tell you if the slow link is actually the SDMB or someone in between. Sometimes an overloaded ISP will drop packets. Comcast actually got themselves in hot water with a lot of their customers when it was discovered that they were intentionally dropping packets and closing connections to ease congestion on their networks.
That was quite a while ago, wasn’t it (if you’re talking about the same incident(s) that I’m thinking of). After that shitstorm, is anybody else still doing that? Or alleged to be doing that?
IIRC, Comcast was doing something considered worse: Inserting a “last-packet” byte into the packets, which told the computers receiving the packet that it was time to close the connection. (I don’t recall if this affected packets from browser to server or from server to browser.) This wasn’t simply dropping packets – they were actually tampering with the data being transmitted. The prevailing attitude, when this was discovered, was that this was a shitstormworthy thing to do.
ETA: Oh, and as for the OP: Yeah, I’ve gotten a few errors like that occasionally. As already mentioned, they seem to be transient.
Minot nit - You’ve got the right basic idea, but what they were actually doing was setting what is called the RST bit in the TCP/IP packet header, which doesn’t add a byte. It does however signal a socket reset, which closes the communication socket between the two computers. Only the computer that is sending the packet should set the RST bit. Anyone forwarding the packet across the internet should never touch it.
For those who don’t understand the technical jargon, it’s basically like when you make a phone call. You call someone else and then talk back and forth. When one of you is done, you hang up. This is basically like Comcast faking it out so that it looks to you like the person you were talking to hung up in the middle of your call. Sure, you can redial and start talking again (which is what computers do - they re-open a new socket and request the data again) but it really slows you down.
Is anyone still doing this? I have no idea. I haven’t heard of anyone getting caught doing it lately.
But packet loss certainly happens, and is usually a sign of an overloaded communication channel somewhere. Since it is something that happens normally when a system gets overloaded, it’s hard to prove if an ISP is doing it to throttle back high usage customers.
An overloaded server (like the SDMB) may also fail to respond to a packet when it gets too busy. In either case you have a packet that wasn’t replied to, so it’s not necessarily easy to tell who dropped the packet, the SDMB or someone in the middle along the way. If you are getting a significantly higher number of errors than the rest of the SDMB users though, that might be a good indication that the problem is somewhere closer to your end.
Timeouts are a lack of a response from the server, which can mean packet loss anywhere along the chain between your computer and the SDMB. Internal server errors and database errors are a response from the server, which means no packet loss but instead indicates an overloaded server.
I’ve run into the following error several times today and don’t recall seeing it before despite regularly getting the database error/server error, etc. messages.
(I love how the messages always make it your problem. My browser sent a complete request in time, buster!)