I thought I would go in to a little more detail and try to explain exactly what is happening. All of the admins agree that what’s happening is very odd indeed. In fact the whole staff team and many others are completely stumped. I know for certain there are a few experts on database management here on the site and we would like to hear some other opinions on what could be causing these strange spikes. We are pretty sure it is to do with innoDB deadlocks and killing connections seems to fix it for a while. Below is an image of the lag spikes taking place. If you would like to give your input you can mail me in game, please only if you have a good experience with this kind of thing. Currently this problem is the ONLY thing causing any lag at all. Finding the source and fixing it means no lag. And there is a source somewhere.
i just sent Ched an in game e-mail. i am a graduate student and currently am working on information systems and focusing mainly on databases and SQL. also my job is as an auditor for a major corporation and audit databases on a regular basis. i think i might be able to help. let me know if you think i can be of service
A bottleneck is exactly what the name implies. It is the point in the system that limits the data throughput, which is the amount of data that can flow through the network
I have a few answers that might fix the problem but i would rather not post onto here as for fear of being criticised but if you do want to hear my suggestions... mail me
P.S. keep up the good work Ched.. this game will always remain the best text based RPG online game EVER.. thanks
In stead of replying with things like "I don't know what you're talking about" or "you'll fix it", don't reply at all. Ched is looking for useful replies here that can help him and the other staff members figure out what's causing these lag spikes. Reading all these pointless replies (including mine, eventually) is just a big waste of time for them.
So do us all a favour and be quiet if you have no useful input on the topic or if you have no idea what this is about.
Ched, I'm not saying I'm an expert, but I know my share of databases and such.
I think that if you want a decent advice, we need more info then just the lag peaks.
If one wants to find out what's causing these peaks, one needs to know what happens in the databases at that moment, to find the relations between them.
For example user-counts? Is there a sudden increase of people logging in? (Maybe the West Coast of the USA gets up at that time for instance)
Certain crons running at that moment (Job payouts, city stats gathering)?
Actually, in short: One should be able to see where all these connections are going to.
I would love to get more in-depth in the problem, but I think one needs more information for that.
Okay, not trying to insult, flame, or attack any individuals, I'm just gonna try and set up a discussion with people who might have some knowledge about the matter.
And maybe that indeed for once....replies such as ":S" could be held on a low?
Mushieman: If it would be a bottleneck problem, that would imply that when there's no lag, the number of connections would be low enough to pass through this bottleneck.
So, the moment the lag occurs, for some reason there is an extreme increase of connections going to the Dbases.
I think one needs to look at that increase. Is it possible that some connections stay open longer then needed? As Ched says, killing all the connections fixes it....for another 24 hours. Are the connections not closed automatically properly, causing a "traffic-jam" of ever increasing number of connections?
If they do shut down correctly automatically, then if it were a bottleneck problem, there has to be some reason to an extreme sudden increase of connections, right?
What happens at that time? Thousands of people logging in at once ( - doubtful)? Crons running at that time? ( - doubtful aswell, since all crons have been taken away)
What is the one remaining thing that occurs only once a day, and generates a lot of connections/calculations?
Shouldn't the answer be searched in that direction?
alot of the lag peaks are around 7pm onwards, about the time alot of people will have got home from work and had dinner ect... could it just be too many of us trying to log in at once and things like that? *confused*
That was one of my thoughts aswell, but disregarded that thougt pretty fast tbh....
If it were simply lots of people signing in, that would have been obvious to anyone monitoring the databases.
Plus.....Ched's new toys (aka the new server set up) should be capable in handling such traffic.
Plus, timezones? I have no clue how the geographical spread is of the players, but I think you'd be seeing more peaks then. One when Europe gets home. one when the US East coast gets home, the West coast, asia etc etc..
@pink they use persistent connections to mysql and you need to see the config files of mysql & apache and the php connection strings to see how the handle the sessions timouts
Guys, if it's related to active connections, how about trying to reduce the inactive timeout.
At the moment the system recognises you as active for 60 minutes after your last action. How about reducing this to 30 minutes?
You have lost your Donater Status. You will no longer recieve 5 energy every 10 minutes and your 150 energy bar has been reduced to 100. If you would like them back, please go here to donate again.
In a word... NO!!!
Guys, if it's related to active connections, how about trying to reduce the inactive timeout.
At the moment the system recognises you as active for 60 minutes after your last action. How about reducing this to 30 minutes?
Follow @TornRPG