We have just purchased a dedicated server. It will be available to us within 2 days, so expect new game servers to be available Wednesday evening.

Also, we would like to perform a stress test on the server to gauge it's performance and what we can run on it. We will try to do this on Saturday, so please try and make yourself available then. Everyone is welcome and we want as many people to join as possible, so invite anyone you want.

More details coming when we get them. Please note that the IPs will not be the same as our existing servers, and those are not the ones we want to test.

This is great news

I should be available for testing. I hope.

Awesomeness!

Just so people know, the donate page should be available when we move stuff over.

Brilliant stuff chaps! Glad to know you made the purchase, hopefully things will move along swiftly.

I'm sure I will be available to push the server to its limits

Perhaps I will have a bit of spare cash for the donations as well. But I'm not exactly swimming in money at the moment

Thanks Goz, if you know anyone else who would help, please ask

UPDATE:

The server is up and running. We are in the process of installing the various things on it, and we will hopefully be able to report to you more about it soon.

Does that mean we can actually use it this evening?

IP?

Depends what time on saturday... But i can stress test on friday or something? During the day i mean

I've been stress testing it and the results are not that promising

Well it seems bots hammer it hard at the beginning (I'm guessing 20 odd bot pathfinders aren't pleasant), but we're not wanting it for bots. Once the bots get to a stopping place the CPU usage drops dramatically from maxing a core to about 50% of a core. I think it will be much better with human players.

Worst case scenario though, right?

20-40odd bots I would consider worst case. Considering we're going to probably to 20 slot publics with no bots (or to make players up to 6, maybe). It hammers the CPU at the round start, but after that it seems to run all right. Well it did when I was on last night.

What is it with people and "its only worst case scenario, it doesnt matter", gauging by what me, DHR and Rannac tested this morning, 25 bots is on the brink of killing it, and the server rejects any more than 41 clients. I never have and probabily never will understand why a justification magically removes a problem, ITS STILL THERE?!?!?!? Ok, 2 servers maybe 6 bots, but if we turn it up on one for one of our normal bot slay matches on awp_india the whole thing is going to die!

Basing on this, i don't see the value for money of this server compared to what we already have.

Also, I send more data to the server about where I am than I receive overall, how is this any good, a server a third of the size (fun server) sends three times more information, I know they are at different tick rates, but not enough to make a damned bit of difference!!! This was measured by looking at the amount of data traveling to and from the IP address.

Yea, I'm sure it might be great when it's done, but for now, it doesn't look too good.

[/rant]

Oh, I'm not saying it doesn't matter, in fact it kind of sucks.

Yeah... it was uber laggy with the amount of bots we had...

I mean it hosted fine, and we all had low latency's but it was laggy as hell with lots of bots...

We simply don't have that many bots then. I was just on and it handled 20bots fine. So if you want bot slay games, just put 12 bots on or whatever, it's not like we have that many now...

I was receiving more data than I was sending, by about 2-3 fold. This was from the net_graph 3 thing.

EDIT: Bit more testing with 40 bots, the server doesn't send much data until about half of the bots or dead, at which point it behaves normally and stops lagging.

The justification removes the problem when the justification is "we don't want to actually do this". When a problem evolves beyond what is expected of normal use, it is not a problem we need to be concerned with. So what, it can't handle 40bots? I could have told you that ages ago, I never intended that we were going to do that.

Then I am starting to question the value for money of this venture seriously now.

2 21slot CSS servers from GameServers = £44/4weeks
2 21 slots CSS servers on this hardware will work = £46/month
The latter is cheaper.

And that is before you factor in flexibility, webserver, teamspeak etc.

Anyway, experience is worth the cost.

Please stop being so pessimistic, it is unnecessary and unconstructive.

Didnt we pay for a 64 slot server tho?

Well f you want to run public servers I can't even understand the fact of why you would want Bots.

Having a public server surely means trying to fill all the slots with humans. Bots are only a nuisance for public servers, people will moan. To me they only ever seem of use for practicting when you have a low number of players.
I.E. I see Wynney on the war server playing with just bots every now and then.

I really can't see how bots would be more useful than a nuisance on public servers.........

No, we paid for a Core 2 Duo 2.2GHz with 1GB RAM and a 160GB HDD. What we do with it is up to us. My aim is to run;

Public CSS server, 20ish slots, 66 or 100 tick
Public TF2 server, 24 slots
Private CSS War server, 11 slots, 100tick (this is mostly unused, won't use CPU until used in wars)
Web server
TS server

From what I have seen so far, this server will run that quite happily. Maybe a bit tight on the RAM, though rebooting seems to free up a lot and it slowly increases in use over time, it seems, so a daily (4am) reboot would probably be set up.
The cost of this from GameServers is (assuming 21slot 100tick public, they don't do 66, and a 15slot discounted TS) is around £50/4weeks + £2/month (for web), or £674/year.
The cost of this server is £46/month £552/year. Even if we can only run that we are still saving £122/year or around £10/month. That's about a 20% saving.
If you throw in the fact that we can do what we want with the dedi, aslong as it's capable, quick SourceTV download section and various other factors, the value is plain to see.
If you make the above public servers private on GameServer's system, GS becomes just cheaper (a matter of a few pounds a year).
If we wish to run more than that, we will probably need a better server, though when we get to the point where we want to run more than for any reason other than "for the sake of it", we should be getting more donations. We could probably run an occasional GMod server or various other things when we want to, though it would be best not to run them permanently and we might need to close one of the other servers to do so.

Please note that when checking my maths, if you so wish, that Gameservers charge on a 4 week billing cycle, meaning 13 cycles/year. This dedicated charges at the first of a calendar month (12/year - the second month is cheaper to bring it in line with the 1st of the month). This is why I calculated the year totals.

Also, Gozzer, you are absolutely correct. I would only want bots on at low numbers (6 or 8 players total) to even up the teams. we are using bots as a stress test as they use more CPU than humans do (and are much easier to test with - it's hard to get 20odd people together on a whim).

No they don't, its monthly, just their invoices come early, I was wrong and I believe I posted rectifying that mistake.

OK, well I wasn't aware of that, and I don't remember such a post.

Still, GS comes out at £6/month more at that, or a total of £624/year, £72 more than what we are trying to do here. Value is still easy to see.

I have edited my previous post to include some more detail in various places, aswell.

Quick update.

Si has been convinced that we might aswell still try it as we've got it for atleast another month ^^

We have TS working properly, just needs configuring by myself (easy task, checkboxes on a webpage mostly). We were considering Ventrillo for improved quality, but the up to date versions only allow 8slots on an unlicenced version (and the licenced version is for resellers only). It's 200slots, but an empty slot doesn't use any resources at all, so we just set a ridiculous number so we won't have to concern ourselves with server full.

Web is working, with PHP and MySQL. However, we are having a problem configuring FTP atm. Aaron has offered to help us later today, and hopefully he will be able to sort it out for us. This simply means we can't upload the site (well we could, but a file by file basis would take forever). When it is done the site will all be moved across, and we do not expect any problems there.

Srcds is all installed and working, though we really need FTP before we can configure them properly, so we are not releasing them yet.

At present we have settled on a 66tick 25slot CSS public server and a 24slot TF2 public server. We figured the extra use of 100tick wasn't particularly worth the extra resources, though it will be subject to testing. Of course, we will also have a 100tick 11slot private CSS server for our wars.

Also, we seem to have discovered what was causing the massive RAM useage. It appears that when we closed srcds, it wasn't closing properly, leaving itself in memory. This meant that every time we wanted to change something such as slot count or tick, it would add about 80MB to the RAM useage. We were doing this a fair bit for experimenting, so it soon added up.

25 slots is more than enough on css, so well done picking a sensible number there.

being honest, there is very little difference in terms of performance between 66 tick and 100 tick (nif is 66 tick and performs v well) so dont increase it i feel, its fine as it is.

could we have a VENT server rather than a TS one pls. vent is higher quality and offers greater flexibility in terms of options.

Sorry, fraid not. I believe TS3 will be improving sound greatly, though.