I don’t know about how servers work, so bear w/ me.
What happens? It just happened for like 10 minutes just now, right? What happens? Does a mod spill a coke onto a hard drive, or fall asleep?
It’s not fun.
Hmph. Kids these days. Why, I remember the day when, when the server went down, the skies went dark and ignorance ruled the land.
Oh, wait.
That was Bill Gates’s fault.
Never mind.
They had to switch out hamsters.
these hampsters today.
why, back in the day hamspters could run all day on one sunflower seed. (well, that and a cat sitting on the opposite counter staring at them…)
Ok, here’s where my mind was…
I hear “server went down”, and my first thought was “Now THAT is what I call a friendly waitress…”
My WAG is that they don’t go down but simply slow down under the strain. The place sometimes gets crowded so while some Dopers are happily typing away, others get booted out, and some can’t get in till someone leaves.
Servers can get into a state where they are spending all their time swapping memory in and out, trashing, and getting nearly no meaningful work done. That state can persist for a while until some processes die, the system crashes, or operator intervention. That could be the situation here; too many users, too many requests, too heavy load, and system starts trashing. It can go on for a while in that state, badly wounded but limping along.
That’s just one possibility, in reality it could be just about anything.
gee - some of the people in my office go thru that too…hmmm
Or a “daemon” (just a funny name for a process, really) decides to take a nap.
Just as an example, if the http daemon quits, gets badly confused or is killed by another process running amok (shouldn’t happen on a *nix server, but there’s always a “but…”) the server will suddenly be unable to serve up web pages or otherwise comminicate with web browsers.
Some servers are configured with “heartbeat” monitors that will periodically poll vital daemons or simply watch their actions and if one stops runing, the misbehaving (or non-behaving!) daemon will be re-started or “bounced” - a process than can take under a second, or perhaps 10-15 minutes if a cold restart is needed.
I go back to doing the work they’re paying me for.
[COLOR=DarkRed]Demons held captive by the thrall of SDMB begin roaming the internet looking for likely victims to rend to and fro.[/COLOR]
Now that’s funny!
So does our ‘server’ computer- is that like, one computer that keeps the whole thing going, and each mod gets the server software for their home pc? Who watches that main computer, and how does it go down and come back up?
Not that post #6 and 7 don’t answer well.
Not knowing specifically how this board is set up, I can only generalize. For a site as busy as this is, I’d like to hope there’s actually two servers - one runs the vBulletin software and handles the transactions - all of the requests to display posts and processing replies or new posts. The other handles the database that stores all of our brilliant words.
We might be on just one physical server, in which case the same computer handles the transactions and stores all of the messages.
The sort of server(s) that would run this operation are outwardly unimpressive unless you’re a geek. The transaction server is probably just under 2" thick, but 19" wide and at least as deep into an equipment rack at the SDMB’s hosting provider. The database server (or a combined server) is probably around 5" tall to hold several drives.
Ultimately, the hosting provider runs the datacenter and manages any hands-on work that might be needed. One of the big buzzwords in servers now is “lights-out” management - it’s all done remotely and nobody even needs to be in the same building. As for making sure it’s running, that falls to the daemons I mentioned earlier. For this, they’re likely to be processes running on a different computer at the datacenter that monitor how well it’s running, or if it’s stopped running entirely, and they’ll intervene as needed - either to “bounce” a process or to completely restart the server.
The moderators are just using regular web browsers to do their jobs - by their ID and password, the system knows who they are and enables moderator functions for them. I administer a far smaller message board, and when I’m logged in as admin, there’s a forest of checkboxes for selecting messages to kill or move and a “control panel” for managing user IDs. In either case, no special software is involved.
The technical term is “thrashing” which is what you propably meant to type. ;j
Jerry once explained it to me this way:
The server performs various tasks in the course of the day, fulfilling requests of various kinds: requests to display pages, requests to write pages, requests for email notification, all the other tasks the server is asked to do to make this board operational.
When the demand/number of tasks is more and faster than the server can deliver them, it does get to a point where it becomes too much. At that point the server drops tasks (this is when you get the “timed out” message) . . . eventually it drops enough tasks that it can reset itself and begin performing tasks again.
This is also why sometimes you will have crappy access and then all of a sudden stuff zips through; enough other requests have been timed out ahead of you that the server can again begin serving and your request got handled. At those times of greatest stress, either your request will be refused or you’ll see blazing speed, most likely.
Some days watching a page resolve on my screen is like watching a fat man climb stairs. <sigh>
TubaDiva
Not only did I mistakenly type “trashing”, I typed it twice. Yes, thrashing is what I meant to say, as I deal with it most days during perf tests. 
Yeah, sometime see some of that too.
Trashing AND thrashing.
Both are not good.
TubaDiva
Wait! So you can see me now?
(Looks around nervously)
Oh well.
(waves!)