EVE Online Dev Tracker

Hello everyone o7

The latest installment of Community Beat is here!

The EVE Online community is vast, and the Beat aims to help pilots discover new parts of the community that they may not have been aware of before. This week we look at a the scientific journey of a Raiju entering the Hypernet, evemoedesu’s structure bash art, some funny highlights from Friendship week which recently concluded and give a little shoutout to the man himfself, Steve Ronuken.

If you have an upcoming event, or thoughts on what you want to see - let us know! It’s getting close to Fanfest and we can’t wait to see you all there.

See ya in a couple of weeks.

This thread is only about the disconnect issues that were affecting Virgin Media UK customers. For your issue, which is of a different nature and with a different ISP, please create a technical support request.

UK status since 1 Mar:

image1140×541 68.4 KB

I’ll update here and on Twitter on a few days with longer-term data, but this case is closed.

You are in Mexico or Chile? If, please file a support ticket.

Could be.

We explicitly suggested to VM on Monday evening that they “try a different egress point to reach 172.65.192.0/20”.

Prior, early in the morning on that same day, we had suggested they could “perhaps change that for 172.65.201.188 (or for Cloudflare in general, AS13335) and route directly within the UK to Cloudflare? Perhaps even initially only as a test to see if a new route makes a difference.”

BTW, the fault was that VM’s systems were trying to establish this direct peering (that they ended up with) but their systems were failing because of prefix limits being breached.

I’m all ears (not literally this time) what that other alternative should have been.

The internet isn’t a single unified entity; it’s a network of networks glued together with BGP and we here at CCP only control our own network.

This is looking good since Virgin Media UK’s fix this afternoon:

image1143×535 60.2 KB

@Lacrimosa: Referring you to my two posts right above, please see

• Right, now I’m interested in hearing if you are all getting new routes?
• Using one of the trace routes from above for some quick analysis

then please note that Virgin Media UK literally fixed the issue themselves by literally fixing

“Issue looks to be due to flapping BGP between Cloudfare and Virgin Media due to prefix limit breach.”

on literally their own equipment, thereby literally fixing/re-establishing their own direct peering with Cloudflare. None of this was or is under CCP’s control or under the control of our internal network partner Advania or our external network partner Cloudflare or any of our network carriers or direct network peers.

I don’t have the details to be able to answer this question – only Virgin Media would be able to tell what the prefix limit was before and after on their equipment, what alerts there are on the limit being breached and their direct peering with Cloudflare therefore being down, how that limit compares to the overall size of the internet’s BGP table and the number of their direct network peers, etc.

Last 10 days; now we wait to see if this graph stabilizes:

image1584×540 110 KB

Hmmm… but your trace route to TQ has changed to the direct peering? I have to admit, I’m slightly concerned.

Using one of the trace routes from above for some quick analysis and details to you, patient EVE players:

3 12 ms 11 ms 10 ms bagu-core-2a-ae70-650.network.virginmedia.net [213.106.214.185]
4 * * * Request timed out.
5 12 ms 16 ms 13 ms tcma-ic-2-ae9-0.network.virginmedia.net [62.253.174.178]
6 22 ms 42 ms 37 ms 162.158.32.254
7 16 ms 16 ms 18 ms 162.158.32.11
8 18 ms 18 ms 14 ms 172.65.201.188

You will note that there is no more Telia in between Virgin Media and Cloudflare, no more Liberty Global, and no more Amsterdam.

Rather in hops 5 an...

Right, now I’m interested in hearing if you are all getting new routes? Virgin Media just told us “Issue looks to be due to flapping BGP between Cloudfare and Virgin Media due to prefix limit breach.”

You should now be getting routes that go directly to the nearest Cloudflare PoP, of which there are 3 in the UK: Edinburgh, Manchester, London – so depending on if you are in the north, south or center of the island then you should get the different PoPs and your route should stay that way during your EVE play session.

EDIT: If Virgin Media is only peering with Cloudflare in one of those three locations then you will all be routed through that location. It looks like Virgin Media is peering with Cloudflare in Manchester.

Actually, @Alaska, can you also post in the ticket the regular trace route, please; for comparison?

Many thanks; forwarded.

@Alaska, can you perhaps send us a support ticket with the full traceroute from hop 1? Please post the ticket number here.

Excellent, we have seen this hop in previous disconnect trace routes but it is not present in the normal ones.

Thanks! Anything that doesn’t look like the regular route, would be helpful to know about.

BTW, Cloudflare was having Cloudflare Status - Network Performance Issues in Amsterdam, Netherlands region this morning. They should have been resolved at 10:31 UTC. Details in the link. (Normally this should not be relevant to UK players, but this entire thread is about strange routing.)