Servers

[Hulk]

Any way we can change this pos server?

[MHz]

We should get a patreon where we can all chuck in $10 a month and get something really powerful. And give Qan a cut to give him some remuneration for all his work.

[qan]

I put in another ticket.

One big possible reason for our trouble has to do with Windows Server 2016 itself, which everyone knows is complete garbage. (Completely ignores policies, acts like a home machine.)


I'd like to consider running our server off a Linux VPS using Wine. However, our hands are somewhat tied because the IP is whitelisted with the billing server, and we're also running a pair of servers (Chaos as well). We'd have to forward the zone ports through to the Linux box, essentially. That's one additional hop, but in the same center so adds very little. Hopefully it'd be reliable.

If the experiment worked and there were no issues, we'd drop the Windows POS entirely and then have the Linux machine take over. But it would be hard to isolate any issues of the Linux machine as related to emulating a server vs related to a misconfiguration of the routing. Not ideal.

Irritating to say the least.

I'm not the only person working on it, either. But we'll add more until we get things figured out. Yeah, it's slow because it's nobody's job, and there are a lot of issues involved on the way as a result.

[qan]

So I had time to do some more investigating: it really was Windows Update once again. I should have guessed. The reason some people could still remain online this time was likely because of the upgrade, though it didn't do that much good. Still was unplayable.

Now, you may ask, how was it updating automatically when updates are disabled?

What a great question!

Windows Server is a crafty devil, possibly one of the worst pieces of software ever written. It switched it back to Automatic after a recent update. Why? YOLO! It can do whatever it wants. (There's a special place in hell for the Windows Server 2016 team.)


Back to manual updates and also disabled quite a few unnecessary services for good measure. Will be checking periodically and after every update to make sure it doesn't turn on automatic updates once again.

[Nipple Nibbler]

This is the sole reason I refuse to upgrade to windows 10, because FUCK auto updates.

[zidane]

lol..

[qan]

Today's lag that ruined the DTJ, meanwhile, wasn't due to high CPU usage associated with updates running any time they damn well please. In this instance we were sending out as much as to 2.5Mbps. That's well under the limit that the server can handle sending at once, but to send it all perfectly reliably without dropping some packets, and additionally while there's a slight uptick in net traffic due to quarantining, is another matter entirely.

We may want to look at cutting back bots a bit during TWDT/TWL especially. We're approaching 250 concurrent players on Sundays and it's pushing the limit of what the server can handle reliably. I did a bit of work this morning to further slim it down, and will continue working on it while consulting experienced spaceships in the community.

If you have an encyclopedic knowledge of Windows Server, particularly regarding the optimization of net traffic, I'd certainly welcome your two cents.

[paradise!]

Can we possibly pay for a consultation from an outside IT expert? I know youre smart af, but could be good to hire a windows server expert who can give us expert advice

[Jessup]

Nikolaos is a wiz on these type matters too. Many players in this game know a ton of stuff about computers and ping and connection issues. Group brainstorm perhaps?

Honestly.. personally my ping and lag been fine since the c2s calmed down last week. at least its no worse than anything in last 10 years I've experienced.

[Ephemeral]

Qan,
Have you looked into WSUS?
Back when I had a rack full of Win servers and domain full of clients I managed all the updating through a WSUS machine. The WSUS server received updates and I scheduled them to be distributed to the network machines. This allowed you to doing testing before the updates rolled out.
NOTE: this was back around Win2003/Win2005 era, things may have changed.
Eph

[Rab]

WSUS still exists, but we wouldn't want to pay the cost for another server. Planning on ASSS/Linux at some point anyway.

[qan]

Just a heads up, I'm on a short TW hiatus due to work demands (which should slow down after a short bit). Community brainstorm is a solid way to go. POiD may be able to fill you in on some details of what's going on, though he's quite busy as well. Pure_Luck also has some awareness of at least how the server was set up.

Ephemeral - Updates should be under control now. We're not seeing massive disconnects as a result, but we're still seeing steady net issues. Provider has not been helpful troubleshooting. If someone would like players to provide MTR reports, maybe we can make use of http://trenchwars.org/lag/ again. Wirah, would you be willing to touch that up slightly? If we can get some traces recorded we can then set up a time to do a two-way trace with those experiencing the worst issues.

[qan]

To add, just did a quick check of trenchwars.org/lag, and the bit about using the in-game trace isn't very useful. We basically need MTR reports. Might be good to remove (or archive at the bottom, something along those lines) to save unnecessary effort

[Iron]

So basically everyday for the past 2 weeks server been fuking us. What's going on? Is there any way we can fix it?

[Beast]

Lol iron complaining when him and skepsis are teleporting experts