CIs, internet issues.

[Borch]

Loanshark was complaining about having CIs and blaming it on our provider. Our provider is located in Atlanta.
Here is a screenshot of the mtr to his IP, which is in California, from the server:

Looking at the image, you can see that 6 and 7 "hops," on the left side, show 56% packet losses, you can also see the pings change there as well. If you lookup those two ips on https://www.iplocation.net/ip-lookup, you will see that they are both from New York, nowhere near Atlanta, where our server is located. Note that these ISPs are a backbone, from Unitas Global, and look like they are going through Boston. Not every hop will show. This might not actually be when he was having trouble, that would require more coordination, doing the mtr test while he monitors his game.

So if you happen to live on the other side of NY from our Atlanta server, especially using a backbone through there, a server in Atlanta or Miami might be terribly slow, while a hosting service based in Chicago might work fine, giving the appearance that our NFO server is defective. Note that they are like us, in that they purchase services from other providers.

I had contacted NFO when we had lots of CIs over the weekend just past and there were no network issues in their bailiwig—consider also that it was reported by Duncan that players on Outlaw were having the same issues, using NFO also.

[Borch]

The only ways to make changes that *might* affect this are to:
1. Move the server within NFO to a new location, such as Chicago (note that there will be complaints from some others about their pings increasing)
2. Move to a different server company (might be a large amount of work).

[bonehill]

Move the server to Dublin please!

[Borch]

I am making a list of companies that I am looking into, what made me not prefer them:

[*]liquid web was discussed on another reviewing site and found them to be charlatans.
[*]Kamatera, 2.7GHz cpus. 1 TB of outgoing Internet traffic
[*]OVHcloud, cpu good, 4.6-4.8 GHz, expensive (50.40?) Rise-LE-1, 4 GHz, 4.9 GHz (turbo?), memory good, storage great, bandwidth is good, only viable location is in Canada (Montreal) Ubuntu server 23.04 (we currently use ubuntu 22.04 LTS), does not look too bad, other than the Montreal server location. Asked for a callback, I might get it for a month to try it out.

[Borch]

This information was provided by NFO:
Re: Attacks on a customer caused some packet loss
Dec 05 2023 12:52:25 PM PT Around 3:30pm CST, we observed heavy packet loss in Atlanta, at the same time as a seemingly small attack against a customer there. Investigating, we determined that it was likely a reflection/amplification attack that was being filtered by our upstream at this location. Unfortunately, measurements showed that the attack was overloading our upstream's links there, causing the packet loss. This may have also occurred several times earlier in the day -- with attacks that looked small on our end because they were being filtered, but which caused upstream loss.

The apparent target was null-routed, but we are watching for more attacks against this customer, who has multiple services. We have also opened a ticket with our upstream to discuss with them what can be tweaked to improve performance against follow-on attacks.

[Borch]

I had difficulty a year ago with mysql being unavailable to B3, so I went around and found some back-door software, but that has prevented software updates along the way, since, so I am resintalling ubuntu on the server and plan to restore the server tonight from backups I made this weekend.

There is a chance I'll screw it up and things won't be working for a while.

[Borch]

My initial post was in error, I had typed the ip incorrectly, and outlaw DOES use NFO, so when they were having problems, it was with them being on the same provider as we.