- Popular Post
Roy
Content Type
Profiles
Articles
Updates
Projects
Twitch
Website Bugs/Suggestions
Guides
Newsletters
About GFL
Knowledge Base
Expenses/Hardware
Server Comparisons
Routing
Form Bugs
Community Representative Applications
Development Request
Forums
Events
Posts posted by Roy
-
-
We are still looking for results from trace routes! Especially since we unannounced IX peers to fix our routing from our physical server to the Dallas PoP.
Thank you!
-
On 3/21/2019 at 4:26 PM, Le Quack said:
Location: Erie, Pennsylvania, US
Try this again. When you performed this trace route, our Dallas PoP was the only one up due to maintenance.
Thanks.
-
- Popular Post
- Popular Post
Hello,
As of March 2019, GFL is now running its own network using Anycast. @Dreae and I started planning this project out in early March of 2019 and as of March 23rd, 2019, the network is running successfully in-front of our game servers on our new dedicated machine. This comes with many benefits such as being able to use our own IPv4 block and with Anycast, this allows us to absorb and filter (D)DoS attacks at the PoP server level which increases overall (D)DoS protection. With us owning the network, we can most of the time influence routing with BGP communities and so on! This means we have a good chance at fixing sub-optimal routes our players get to our PoP servers. We are still exploring the world of BGP routing and there's a chance we'll find better methods to correct sub-optimal routes, though.
We have leased a /24 IPv4 block (256 IPs) and have ownership over this block. The IP block is 92.119.148.0/24. The good thing about owning an IPv4 block is we will own it as long as we pay for it (the block itself is only $56.00/m). If we need to switch physical hosting providers, we can do so without an IP change since we control which hosting providers use this IP via BGP. With that said, our Anycast network forwards traffic from each PoP server (Point Of Presence) to the physical hosting provider itself using @Dreae's SRCDS packet filtering system on Linux which is open-source and can be found here. Therefore, we don't technically need a physical hosting provider that supports BGP sessions unless if we ever get rid of the Anycast network itself. Thankfully, our new physical hosting provider supports BGP sessions either way. Therefore, if we were to remove our Anycast network, we will be able to use our IPv4 block with them directly.
This is by far the best project I've ever worked on personally. I haven't seen any other standard gaming community literally own their network like we are doing. With the said, @Dreae and I have been learning a lot from this project and there is so much more to learn! This is definitely helping further advance my networking knowledge. We will be posting a public guide at some point outlining how we did all of this for under $200/m along with some advice/tips we found on the way. No ETA on when this guide will be released publicly.
With the amount of on-going changes with routing and networking, this network will need active monitoring and maintenance most likely. This is why we're having a form where clients can post sub-optimal routes they find via trace route/MTR to our PoP servers.
All in all, this is a big addition to GFL. I also want to thank @Dreae for all he has done for this project! He has put many hours into creating the packet filtering software running on our PoPs, giving me advice on the network setup in general, and helping with BGP routing/BIRD configuration.
If you have any questions, feel free to reply.
Thanks!
-
Just a heads up, every PoP server besides Dallas is offline at the moment (the BIRD daemon isn't running). We are trying to fix the route from our physical machine to the Dallas PoP and experiencing issues. Once we fix this, we'll put all the rest of the PoPs back up.
This means every trace route performed will route to the Dallas PoP regardless of location, etc.
Thank you.
-
The machine appears to be online again:
C:\Users\Christian Deacon>ping 185.177.22.161 Pinging 185.177.22.161 with 32 bytes of data: Reply from 185.177.22.161: bytes=32 time=130ms TTL=43 Reply from 185.177.22.161: bytes=32 time=222ms TTL=43 Reply from 185.177.22.161: bytes=32 time=130ms TTL=43 Reply from 185.177.22.161: bytes=32 time=242ms TTL=43 Ping statistics for 185.177.22.161: Packets: Sent = 4, Received = 4, Lost = 0 (0% loss), Approximate round trip times in milli-seconds: Minimum = 130ms, Maximum = 242ms, Average = 181msI have restarted the game servers on the machine and requested a reason for outage from our hosting provider.
Thank you.
-
Hello,
Currently, our GS05 machine located in London, UK is currently down. As of right now, we have no information on why it's down, but our hosting provider is investigating.
All we know is this:
[ChristianDeacon.DESKTOP] ➤ ping gs05.gflclan.com Pinging gs05.gflclan.com [185.177.22.161] with 32 bytes of data: Reply from 185.177.22.161: Destination host unreachable. Reply from 185.177.22.161: Destination host unreachable. Reply from 185.177.22.161: Destination host unreachable. Reply from 185.177.22.161: Destination host unreachable. Ping statistics for 185.177.22.161: Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),I will reply back here once I have an update.
I apologize for the inconvenience and thank you for understanding.
-
1 hour ago, King_Wailord said:
Location: Zionsville, Indiana
Traceroute:
Tracing route to 92.119.148.5 over a maximum of 30 hops
1 3 ms 2 ms 4 ms testwifi.here [192.168.86.1]
2 3 ms 2 ms 4 ms 192.168.0.1
3 9 ms 6 ms 8 ms cytninhed11-dy1001-secondary4.network.tds.net [96.60.208.1]
4 9 ms 10 ms 10 ms cytnindst52-tg0-7-0-0.network.tds.net [64.50.231.113]
5 31 ms 30 ms 32 ms h64-50-224-143.mdsnwi.tisp.static.tds.net [64.50.224.143]
6 35 ms 30 ms 31 ms nyiix.gi3-6.cr1.nyc1.choopa.net [198.32.160.157]
7 29 ms 30 ms 34 ms vl42-br2.pnj1.choopa.net [108.61.2.89]
8 * * * Request timed out.
9 * * * Request timed out.
10 * * * Request timed out.
11 * * * Request timed out.
12 * * * Request timed out.
13 * * * Request timed out.
14 * * * Request timed out.
15 * * * Request timed out.
16 * * * Request timed out.
17 * * * Request timed out.
18 * * * Request timed out.
19 * * * Request timed out.
20 * * * Request timed out.
21 * * * Request timed out.
22 * * * Request timed out.
23 * * * Request timed out.
24 * * * Request timed out.
25 * * * Request timed out.
26 * * * Request timed out.
27 * * * Request timed out.
28 * * * Request timed out.
29 * * * Request timed out.
30 * * * Request timed out.Trace complete.
You'd probably be better off being routed to the Chicago PoP. I can look into this later.
Thanks.
-
5 minutes ago, RainGamma said:
I have one gig ethernet.
I hate to break it to ya, but we don't even have a PoP in Virginia and I'm not sure how you're even getting a reply from the Anycast IP itself.
I think your ISP is doing something fucky, lol. I'm just not sure what. Still though, 9ms is pretty good
You're probably getting routed to the NYC PoP.
Thanks.
-
8 minutes ago, RainGamma said:
Location - Sterling, VA
Trace Route -
Tracing route to 92.119.148.5 over a maximum of 30 hops1 <1 ms <1 ms <1 ms 192.168.1.1
2 9 ms 11 ms 7 ms 92.119.148.5Trace complete.
That is the best trace route I've seen lmao.
I honestly don't even know how you got a reply from the IP itself, lol.
Thanks.
-
28 minutes ago, SwegBuster said:
I do actually get routed to Seattle before going to LA. The others are closer to my state so.
Thanks.
Yes, but it routes you from a hop in Seattle all the way to the PoP we have in Los Angeles (this would just add latency since you're being routed to Seattle and back to Los Angeles) instead of just going to the PoP we have in Seattle. This is most likely due to the route to Los Angeles having less hops than the route to the Seattle PoP (Anycast picks the closest PoP based off of the amount of hops it takes to get to each PoP I believe). This leads to a routing issue. I may PM you asking if you can perform a trace route to our Seattle PoP server so I can see what route you're taking to there and then make optimizations where necessary.
I will admit though, if our physical server is in Dallas, you're better off being routed to the Los Angeles PoP. However, the Seattle PoP is still technically closer. Therefore, there is still optimizing we have to perform with the routing.
I've noticed our Seattle and Dallas PoPs having the least amount of traffic and these are the two PoPs we have the most routing issues with. Once I get more experience with BIRD/BGP configuration, myself or @Dreae (he is definitely more experienced with this) should be able to correct this.
I am trying to find out how advanced/specific the BIRD configuration can go. It'd be really cool if we could optimize certain routes based off of the destination IP. For example, the IPs that get forwarded to the physical server in Dallas, it'd be better to prefer you to the Los Angeles PoP but if a physical server is in Chicago, you'd be better off getting routed to the Seattle PoP. I'm not sure if this is even possible, though.
Thanks!
-
@SwegBuster @Bae Hm. Both of you should be routed to a different PoP. Swegbuster should be routed to the PoP in Seattle, WA and Bae should be routed to the PoP in Dallas.
I will investigate this later on and see what BGP/BIRD filters I can change.
Thanks.
-
1 hour ago, Fafy said:
Location: Argentina, Buenos Aires (Province), La Plata (City)
C:\Users\usuario>tracert 92.119.148.5 Traza a 92.119.148.5 sobre caminos de 30 saltos como máximo. 1 <1 ms <1 ms <1 ms 192.168.1.1 [192.168.1.1] 2 3 ms 2 ms 2 ms 10.100.100.1 [10.100.100.1] 3 3 ms 3 ms 2 ms 143.208.36.241 4 5 ms 4 ms 4 ms static1.100.12.190.cps.com.ar [190.12.100.1] 5 4 ms 4 ms 4 ms static.40.192.104.190.cps.com.ar [190.104.192.40 ] 6 5 ms 4 ms 4 ms 195.22.220.113 7 34 ms 33 ms 31 ms et11-1-0.sanpaolo8.spa.seabone.net [195.22.219.1 45] 8 141 ms 141 ms 142 ms ntt-verio.sanpaolo8.spa.seabone.net [149.3.181.6 5] 9 141 ms 140 ms 141 ms ae-8.r24.nycmny01.us.bb.gin.ntt.net [129.250.2.1 2] 10 141 ms 141 ms 140 ms ae-1.r07.nycmny01.us.bb.gin.ntt.net [129.250.3.1 81] 11 141 ms 141 ms 140 ms ae-0.a02.nycmny01.us.bb.gin.ntt.net [129.250.6.5 1] 12 150 ms 141 ms 173 ms ae-1.choopa.nycmny01.us.bb.gin.ntt.net [128.241. 2.250] 13 141 ms 142 ms 141 ms vl42-br2.pnj1.choopa.net [108.61.2.89] 14 * * * Tiempo de espera agotado para esta solicitud. 15 * * * Tiempo de espera agotado para esta solicitud. 16 * * * Tiempo de espera agotado para esta solicitud. 17 * * * Tiempo de espera agotado para esta solicitud. 18 * * * Tiempo de espera agotado para esta solicitud. 19 * * * Tiempo de espera agotado para esta solicitud. 20 * * * Tiempo de espera agotado para esta solicitud. 21 * * * Tiempo de espera agotado para esta solicitud. 22 * * * Tiempo de espera agotado para esta solicitud. 23 * * * Tiempo de espera agotado para esta solicitud. 24 * * * Tiempo de espera agotado para esta solicitud. 25 * * * Tiempo de espera agotado para esta solicitud. 26 * * * Tiempo de espera agotado para esta solicitud. 27 * * * Tiempo de espera agotado para esta solicitud. 28 * * * Tiempo de espera agotado para esta solicitud. 29 * * * Tiempo de espera agotado para esta solicitud. 30 * * * Tiempo de espera agotado para esta solicitud. Traza completa.
Interesting. I plan to at some point get PoPs in South America. But for some reason you're being routed to the NYC PoP instead of something like Miami (preferred based on your location) or Dallas.
If you can, would you be able to PM me your public IPv4 address? I'd like to investigate this further. I may ask to do trace routes to specific PoPs themselves if that's okay.
Thank you.
-
6 minutes ago, Liloz01 said:
First - A college computer
Second - My laptop running off of my mobile hotspot (I've played games like this before, don't judge)
The first trace route being a college computer makes sense. I think your college may be blocking that IP since it doesn't even leave the internal network, lol.
The high latency on the second trace route now makes sense since you're using a mobile hot spot haha.
Thanks!
-
22 minutes ago, Liloz01 said:
What devices/networks are being used in the first two trace routes? The first one doesn't even leave the internal network and the second one has ~70ms on just the second hop.
The third one looks pretty good though
Thanks!
-
I just wanted to say thanks again for the results! In the future, we are most likely going to spin up a PoP server in Atlanta, US for players closer to there. However, other than that, things are looking great! We will need to make some routing changes for our physical server's hosting provider, though.
GFL having its own network is very exciting and I really haven't seen any other standard gaming community do that yet.
As stated in the original post, I will be making a more detailed announcement listing what this exactly does and how it benefits GFL once the network is completely rolled out.
Thank you!
-
16 minutes ago, Remuchu said:
Route to Dallas from Louisiana is down I believe, cant trace it and other locals cant connect at the moment
The timeouts at the end are normal since there isn't anything replying to the request via ICMP. The last hop that sends a reply is usually where you end up at. Most of the results route to Chicago. @Ben and I route to Dallas.
EDIT
@nblockbuster routes to the NYC PoP.
Thanks.
-
2 minutes ago, Ben said:
Location - Pickens, S.C., US
Tracert -
Tracing route to 92.119.148.55 over a maximum of 30 hops 1 6 ms 9 ms 9 ms 192.168.0.1 2 * * * Request timed out. 3 29 ms 29 ms 29 ms 10.2.1.21 4 27 ms 19 ms 19 ms 4.34.18.21 5 * 39 ms * ae-1-51.ear3.atlanta2.level3.net [4.69.150.225] 6 33 ms * * atl-b22-link.telia.net [213.248.84.34] 7 39 ms 31 ms 37 ms atl-b22-link.telia.net [62.115.113.24] 8 71 ms 57 ms 49 ms dls-b22-link.telia.net [62.115.120.113] 9 58 ms 59 ms 49 ms vultr-ic-318447-dls-b22.c.telia.net [62.115.149.45] 10 * * * Request timed out. 11 * * * Request timed out. 12 * * * Request timed out. 13 * * * Request timed out. 14 * * * Request timed out. 15 * * * Request timed out. 16 * * * Request timed out. 17 * * * Request timed out. 18 * * * Request timed out. 19 * * * Request timed out. 20 * * * Request timed out. 21 * * * Request timed out. 22 * * * Request timed out. 23 * * * Request timed out. 24 * * * Request timed out. 25 * * * Request timed out. 26 * * * Request timed out. 27 * * * Request timed out. 28 * * * Request timed out. 29 * * * Request timed out. 30 * * * Request timed out. Trace complete.
Looks like you're getting routed to the Dallas PoP. I'm wondering if you'd be better off getting routed to the NYC PoP. I guess it depends on where the physical server is hosted (if it's overseas, NYC is better, if it's in Dallas, TX, that PoP would obviously be better).
Thank you for the results so far everyone! I'm liking the routes and no concerns as of right now
Thanks!
-
- Popular Post
- Popular Post
Hi all,
@Dreae and I will be deploying a new project (we called it 'Operation Turbofan', lol) in the next few days which will highly benefit GFL in the near future. A more detailed announcement will be coming once this project is fully completed and rolled out (within the next week more than likely).
Anyways, in short, we're rolling out our new Anycast network (which we own) that will be in-front of future game servers we host. With that said, we own an IPv4 /24 (256 IPs) block which is 92.119.148.0/24.
The hosting provider we're using to host our PoP servers at the moment allows us to influence our routing with BGP communities through BIRD. However, we'd like to see what routes our players are taking to our Anycast IPs. This will allow us to make tweaks to our routing through BGP/BIRD.
We have PoP servers in the following locations and plan to introduce new PoP servers as time goes on (especially once we acquire our own ASN which is an ongoing process right now):
- Dallas, TX, US.
- Chicago, IL, US.
- New York City, NY, US.
- Miami, FL, US.
- Los Angeles, CA, US.
- Seattle, WA, US.
- Frankfurt, DE.
- Paris, FR.
- London, UK.
In order to see the route you're taking, you can perform a trace route or MTR to this IP range 92.119.148.1 - 92.119.148.254. For example, you can use 92.119.148.5.
In Windows, you can perform a trace route by opening "Command Prompt" and typing:
tracert 92.119.148.5
Afterwards, highlight the output and right-click (this will copy the highlighted text). You can then paste it in the format below.
Please use the following format:
Location - <Your location, be as specific as you can be>
Trace Route -
Output of the trace route here.
For example:
Location - San Antonio, TX, US
Trace Route -
C:\Users\Christian Deacon>tracert 92.119.148.6 Tracing route to 92.119.148.6 over a maximum of 30 hops 1 <1 ms <1 ms <1 ms 192.168.11.1 2 22 ms 18 ms 17 ms cpe-24-243-32-1.satx.res.rr.com [24.243.32.1] 3 33 ms 25 ms 33 ms tge0-0-4.lvoktxad02h.texas.rr.com [24.28.133.245] 4 23 ms 13 ms 13 ms agg20.lvoktxad02r.texas.rr.com [24.175.33.28] 5 19 ms 23 ms 13 ms agg21.snantxvy01r.texas.rr.com [24.175.32.152] 6 180 ms 63 ms 58 ms agg23.dllatxl301r.texas.rr.com [24.175.32.146] 7 28 ms 30 ms 31 ms 66.109.1.216 8 118 ms 72 ms 22 ms 66.109.5.121 9 90 ms 80 ms 73 ms dls-b21-link.telia.net [62.115.156.208] 10 22 ms 27 ms 29 ms dls-b21-link.telia.net [62.115.113.85] 11 25 ms 30 ms 22 ms dls-b22-link.telia.net [62.115.137.107] 12 36 ms 25 ms 20 ms vultr-ic-318447-dls-b22.c.telia.net [62.115.149.45] 13 * * * Request timed out. 14 * * * Request timed out. 15 * * * Request timed out. 16 * * * Request timed out. 17 * * * Request timed out. 18 * * * Request timed out. 19 * * * Request timed out. 20 * * * Request timed out. 21 * * * Request timed out. 22 * * * Request timed out. 23 * * * Request timed out. 24 * * * Request timed out. 25 * * * Request timed out. 26 * * * Request timed out. 27 * * * Request timed out. 28 * * * Request timed out. 29 * * * Request timed out. 30 * * * Request timed out. Trace complete.
Mine routes to the Dallas, TX PoP. However, we needed to do some tweaking in the BIRD config because at first, I was being routed to the NYC PoP instead of Dallas.
If you need any help, please let me know!
You may PM @Dreae or I the trace route as well if you don't feel comfortable posting it publicly.
Thank you for your time.
-
This is something we can look into after deploying our network (which will happen in the next week or so).
Thanks.
-
I have some YouTube videos I made at the time CS:S ZE started getting really popular:
This was back in 2011:
I have many other videos of GFL's servers back in 2011 - 2012 on my YouTube channel.
Thanks!
-
- Popular Post
- Popular Post
I do certainly feel getting involved with eSports in the future will be beneficial to us. Though, I do think we need to ensure we have a more stable structure once we do so again. I don't think our structure was stable last time we attempted an eSports division which is why we don't have it now.
I don't see this being a thing any time soon due to the big projects we're working on now which should be getting rolled out in the next couple weeks or so (exciting news will come soon). But meh, who knows.
Thanks.
-
Fixed
- Performance issues when attempting to post. Sometimes it'd take up to a minute to post a thread depending on the size.
The cause was due to Google's DNS server being somehow blocked by our firewall. We were using Google DNS beforehand on our machines. My guess is when making a post, the server made some DNS look-ups, but since the DNS server was blocked, it was timing out. This not only caused issues on our main website, but also with other services that ran on our web machine.
Since then I've white-listed the Google DNS server on our firewall and I don't expect to run into this issue again.
Thanks.
-
-
On 2/25/2019 at 4:51 PM, Bara said:
Just my opinion:
Is a older non-overclockable 3.50ghz/3,90ghz(boost) cpu with 6 cores (12 threads with hyper threading). It doesn't make sense (for me) to switch to this kind of server, because the 4790k has more single core performance with less cores/threads.
If we were to purchase this machine, we wouldn't be putting our higher-slot servers on it. Instead, it'd be primarily used to host our mid-range game servers and even out the load on our current machine in Dallas (we would not get rid of the Dallas machine with GMCHosting).
We still plan to purchase the Intel i7-9700K once it becomes available. However, at this point, it's sounding like it won't be available for a while.
Thanks.

[Announcement] [MAJOR] Our New Network, Physical Dedicated Machines + More!
in Announcements
Posted
Hello everyone,
I would like to announce the release of a major project that we’ve been working for the last three weeks or so. GFL now owns its network that will be running in-front of our future game servers! This is by far my most favorite project I’ve worked on, ever. While the last three weeks have been very hectic (exchanging many emails/tickets/paperwork, actively trying to learn BGP routing + BIRD configuration, and so on), I do believe the end result will definitely be worth it, especially for GFL! I can say personally that I am learning a lot from this project and it further advances my knowledge in networking as a whole.
Our New Anycast Network
@Dreae and I decided to build an Anycast network to put in-front of GFL’s game servers. For those that do not know what an Anycast network is, this article by CloudFlare explains it well in my opinion.
Anycast itself will benefit us because we’ll be able to absorb and filter (D)DoS attacks at the PoP server level. This results in overall better (D)DoS protection. With that said, the more PoP servers we have, the more network capacity we’ll contain.
Our Current Setup
As of right now, we have nine PoP (Point Of Presence) servers scattered around the world and plan to add more as time goes on. These locations include:
Here’s our Anycast layout. As of right now, we’re using the Vultr setup and don’t have any PoP servers in Tokyo and Singapore. Though, we plan to get PoP servers in these locations in the future. As of right now, the above PoP servers combined cost us $135.00/m.
Our PoP servers run SRCDS packet filtering software made by @dreae which is open-source and can be found here. @Dreae has done a great job with this!
Ongoing Maintenance
Due to the nature of routing and networking, this new Anycast network will more than likely require active monitoring for sub-optimal routes due to things always changing. This is why we have made the following sub-forum:
https://gflclan.com/forums/forum/959-gfls-network/
In here, we will make a form where users can submit sub-optimal routes they find to our PoP servers via traceroute/MTR. With that said, we will be posting maintenance and outage events in here as well.
@Dreae and I are still exploring the world of BGP routing and BIRD configuration. However, we are making progress by using BGP communities and so on to influence routing to our PoP servers. I hope we can eventually find something better than BGP communities, but it’ll do for now.
F.A.Q.
Q - Do you plan on making a public guide going over how you’ve done all of this?
A - Yes, @Dreae and I plan to be as transparent as possible. We will be making a public guide in the future explaining everything we had to go through in order to create this network. We are still pretty new to all of this ourselves, but as time goes on and we learn, we will be making our techniques/advice public!
Q - Would you be up to host non-GFL servers via the Anycast network?
A - As of right now, this will be used specifically for GFL only. However, I am considering this in the far future (I’d LOVE to become a hosting provider some day, but A LOT goes into that and I need to gain more knowledge in networking itself before that’s a possibility).
Q - Do you plan to use other providers to host your PoP servers?
A - Yes, this is the plan and overall, this would make our network better since we wouldn’t be relying on one hosting provider to host our PoP servers. However, in order to do this, we will need our own ASN for our IPv4 block. This is a longer process since it requires verification, contracts with at least two transit providers, etc. However, in the next week I will be looking to apply for one through RIPE (our IPv4 blocker provider can sponsor us and we have two transit providers as well).
Our New IPv4 Block
We have leased our very own /24 IPv4 block. This comes with 256 IPs as Anycast only supports a minimum of /24 IPv4 blocks. Thankfully, we were able to get this entire IPv4 block for only $56.00/m, which in my opinion, is a very great deal!
The IPv4 block is 92.119.148.0/24 (usable IPs go from 92.119.148.2 to 92.119.148.254).
While being able to use this with our Anycast network is a benefit of its own. There are also other benefits of owning an IPv4 block. For example, if we were to want to switch physical hosting providers, we can do so without any IP changes as long as it’s a BGP-supported hosting provider. However, since our Anycast network itself forwards traffic from the PoP servers to the physical hosting provider, we won’t need a BGP-supported physical hosting provider. Thankfully, our new physical hosting provider has BGP-support either way. Therefore, if we were to remove our Anycast network for whatever reason, we can still setup a BGP session with them which will result in our IPs still not changing.
Geo Location
Since this IPv4 block was purchased using RIPE, the geo-location of the block was either non-existent or pointing towards somewhere in Europe. We want this to be set to Dallas, TX, US to start. I’ve emailed three of the biggest geo-location database providers (MaxMind, IP2Location, and Neustar). MaxMind has already made this correction last week. I emailed IP2Location last night and they replied wanting further verification which I then provided. Therefore, I am just waiting to hear back from them. Neustar has yet to reply.
I believe the Valve Master Server may use IP2Location (I initially thought they used MaxMind, but I believe I’m wrong). Therefore, our game servers under this IPv4 block may show up late or incorrectly in the Valve server browser based off of the client’s geo location. Once the correction is made through IP2Location, we’ll just have to wait for the Master Server to update their database. As soon as that’s done, we should be good to go!
Acquiring Our Own ASN
The next big process is acquiring our own ASN. As of right now, we have to use a hosting provider’s public ASN (in this case, Vultr’s) since we don’t have our own. Once this is done, we’ll be able to use multiple hosting providers for our PoP servers. This will make our network more stable/better and less costly since we’ll be able to use cheaper and more efficient providers with lesser locations.
Our IPv4 block provider will be able to sponsor us. Therefore, the only other thing I will need is contracts with two transit providers (I believe I have two I can use) and my driver’s license. After I apply for our own ASN, the process itself will take around 2 - 3 weeks.
New Physical Hosting Provider In US
In addition to the above, we have found a new physical hosting provider that will be ran under the new network. We have purchased two machines from this new hosting provider which are the following:
First
Second
I was aware that our GS06 machine was very overloaded. Therefore, I decided to purchase two machines from the start to be on the safe side. This hosting provider will be able to setup machines within a reasonable time-frame as well and appears to always have dedicated machines available on the spot.
The first machine will host medium-ranged servers (e.g. <40 slot servers). The second machine will be used to host our bigger servers.
We found that this hosting provider also has pretty decent routing and good support. I believe they’ll work for us in the future! In the end, if they don’t, we’ll be able to switch hosting providers without changing IPs
What Servers Will We Be Running?
With the new physical machines, we will have a lot of room to play with. We will be moving servers over to this new network that aren’t so popular. We will primarily move servers in Garry’s Mod and Rust. Unfortunately, any moved servers will require an IP change. But I’m very hopeful that’ll be the last time we switch IPs
We also plan to setup new servers as well! As soon as our IPv4 block’s geo location is corrected, I believe we’ll be good to go.
As of right now, I don’t have any list of servers we’ll be moving. This will be up to the Division Leaders + Directors. I will say I would like to see our Garry’s Mod servers from GS06 moved along with Rust.
Test Servers
I’ve setup some silly test servers on our first new physical machine. You can find all servers that run under our new IPv4 block on GameTracker here.
Feel free to connect to these servers and see how your connection is from the PoP servers to the new physical hosting provider! Remember, you will be able to submit sub-optimal routes if you have one under the new network forum found here.
Special Thanks To @Dreae!
I just wanted to thank @Dreae again! He has been a big help with this project. He has made the packet filtering software that runs on all of our PoPs, given advice on the network setup itself, altered some needed routing via BGP and BIRD, and more! This project definitely wouldn’t be possible without him.
Conclusion
All in all, this is a BIG addition to GFL. There are going to be required tweaks in the future and there is still a lot for @Dreae and I to learn. But I am confident we’ll do great!
I have yet to see another gaming community that hosts game servers have their own network that runs in-front of their game servers. I believe this will be something special to GFL. I’m hoping the public guides @Dreae and I post publicly will influence other gaming communities to do the same since in my opinion, this is fairly cheap to setup (the entire setup itself is costing us <$200.00/m and we will be able to get that cost down once we acquire our own ASN). It might be hectic at first, but if you’re interested in networking, you’ll definitely enjoy and learn a lot from it
You’re basically becoming your own mini-ISP!
If you have any questions, please reply to the thread or post in the GFL Network sub-forum!
Google Doc
Thank you for reading!
View full article