Catalyst Gaming
General => Suggestions => Topic started by: Tyrex on February 07, 2012, 09:06:36 PM
-
OA and its unpredictability causes many server crashes, way too frequently. Instead of addressing OA's crashing problem, it would be much easier to just set up a server auto-restarter. I would not be as annoyed as I am when the server spontaneously crashes, as long as it restarts soon after. This next day thing is just atrocious. I'm not sure if Rofl directly hosts the servers himself, or has someone else do it... But I was searching around and I discovered this: http://tinyurl.com/7fsneps . ServerChecker on the other hand communicates directly with the server the same way a player would, so if the server doesn't respond to user queries it will get restarted. On top of that, ServerChecker is completely free.
TL;DR: Server Auto-Restarter Instead of Waiting
BTW, sorry for the raging.
-Tyler
-
Yeah this would be a good idea. But I don't know what the technical side of what they have going looks like.
-
Not sure how this would work for us due to how our backend works, so I'll let Somone77 answer this one.
-
The fact that the status bar already has an option to check if the server is frozen, and the admins has an option to restart servers (panel is currently broken but the system still works i believe), this should be pretty easy for Somone77 to do. On the box, there is a check that updates the status panel, extend that to restart a server if its frozen for x amount of seconds.
Could make a check and set the value to the value + 1, and if that reaches 5 or something, it would restart.
-
Personally, I would set it for a minute or two, if its like 5-10 seconds, a simple bit o lag would reset the server, or a 3 second shellbooter for example, same thing.
-
Personally, I would set it for a minute or two, if its like 5-10 seconds, a simple bit o lag would reset the server, or a 3 second shellbooter for example, same thing.
The status bar is updated every one minute, so a five value would mean five minutes for a restart..
-
Unless this has a proper interface, I would have to say no.
-
Unless this has a proper interface, I would have to say no.
Just because it doesn't have a stupid friendly interface doesn't mean it wont work ya know.
Still would be worth a try IMO. May actually lead to more uptime and Graphics cards donations for rofl.
-
Unless this has a proper interface, I would have to say no.
Just because it doesn't have a stupid friendly interface doesn't mean it wont work ya know.
Still would be worth a try IMO. May actually lead to more uptime and Graphics cards donations for rofl.
^ That. I don't think a proper cure will ever come out for preventing the crashes as how much shit is piled into OA. However, reducing the downtime would also increase uptime, therefore promoting the server even further.
-
In essence, the back end is already there, and it would take me less than a second for me to do it, but lets look at what you guys are actually requesting here.
The tool posted above works the same exact way as my status bar does. It, first, pings to server to test its connection, then, connects via steam queries to verify that it's processing. If it's connecting, but not processing, it resturns "Frozen".
I've thought many times about having the server auto restart when detected as frozen, however, there is a huge chance of false detection. Since the updater runs once every minute, if, at that EXACT minute, the server lags for a second (you know that red connection text on your screen? That would cause this.) it would restart, even if it was just a small lag. This is why I implemented the server management page.
The management page allowed admins to restart servers if they're down, this means that a person can verify that it's down, not just lag.
tl;dr it already works, it's just RTLK broke the forum and you'll have to wait while we fix shit.
-
Make it so it only restarts if it sense it's frozen twice in a row?
-
It's hard to see if you don't watch the updater while it's running, but false detection is extremely common. You guys would think this is nice, until the day that it restarts when you're growing weed or whatever you guys do.
Trust me when I say: Let me admins restart them.
-
Let me admins restart them.
It's fine if the admins restart them... But I just don't see why it takes a day to restart the server.
-
Let me admins restart them.
It's fine if the admins restart them... But I just don't see why it takes a day to restart the server.
Did you even read what he said? The server management page is temporarily broken due to issues.
-
Also, the problem might not be just a restart. If a server is unable to connect to a database, it will not work just because of a restart.
-
Also, the problem might not be just a restart. If a server is unable to connect to a database, it will not work just because of a restart.
Code something smarter than your problems. IE Something that detects Database up/downs and scans it thrice before actually committing the restart.
-
I don't know if you would understand this, but there are a near infinite and unpredictable amount of things that go wrong, especially with our servers. If I were to take your advice, I would have to make the updater, for every server, run mysql queries, server queries, a2s pings, profile requests and dozens of other checks. This would take near a minute for each server which causes more and more problems.
I'm going to lock this as we've internally discussed this topic to death. We're not going to have servers auto restart because there's always that chance of a false detection. You all would hate auto restarts and you would realize that the second it restarts and loses you money.