Cool Ideas Fibre ISP – Feedback Thread 4

DrJohnZoidberg

Honorary Master
Joined
Jul 24, 2006
Messages
22,699
What's the state of IPv6 currently? I had it working with WebSquad but had to disable it because there was something wrong causing issues.
 

leppie

Senior Member
Joined
Mar 25, 2019
Messages
527
Jirre these Foctotel disconnects every 1-2 days are working on my tits now. This is NOT acceptable anymore.

Also got JHB routing to 1.1.1.1 somehow again.... @TheRoDent my IP is 102.132.200.72

Tracing route to one.one.one.one [1.1.1.1]
over a maximum of 30 hops:

1 <1 ms <1 ms 2 ms ZyXEL.Home [192.168.0.1]
2 5 ms 3 ms 2 ms 154.0.0.238
3 2 ms 2 ms 2 ms 100.98.0.2
4 7 ms 7 ms 5 ms 100.98.0.4
5 21 ms 21 ms 22 ms cloudflare.ixp.capetown [196.60.70.198]
6 18 ms 18 ms 18 ms one.one.one.one [1.1.1.1]

Trace complete.

Edit: Fine again after reconnect. New IP is 155.93.247.8x
 

PBCool

Cool Ideas
Company Rep
Joined
Jan 11, 2016
Messages
11,084
Jirre these Foctotel disconnects every 1-2 days are working on my tits now. This is NOT acceptable anymore.

Also got JHB routing to 1.1.1.1 somehow again.... @TheRoDent my IP is 102.132.200.72

Tracing route to one.one.one.one [1.1.1.1]
over a maximum of 30 hops:

1 2 5 ms 3 ms 2 ms 154.0.0.238
3 2 ms 2 ms 2 ms 100.98.0.2
4 7 ms 7 ms 5 ms 100.98.0.4
5 21 ms 21 ms 22 ms cloudflare.ixp.capetown [196.60.70.198]
6 18 ms 18 ms 18 ms one.one.one.one [1.1.1.1]

Trace complete.

Edit: Fine again after reconnect. New IP is 155.93.247.8x
Apologies I will try get the latest update for you. Re cloudflare drop me a PM if it happens again.
 

TheRoDent

Cool Ideas Rep
Joined
Aug 6, 2003
Messages
5,695
Jirre these Foctotel disconnects every 1-2 days are working on my tits now. This is NOT acceptable anymore.

Also got JHB routing to 1.1.1.1 somehow again.... @TheRoDent my IP is 102.132.200.72

Tracing route to one.one.one.one [1.1.1.1]
over a maximum of 30 hops:

1 <1 ms <1 ms 2 ms ZyXEL.Home [192.168.0.1]
2 5 ms 3 ms 2 ms 154.0.0.238
3 2 ms 2 ms 2 ms 100.98.0.2
4 7 ms 7 ms 5 ms 100.98.0.4
5 21 ms 21 ms 22 ms cloudflare.ixp.capetown [196.60.70.198]
6 18 ms 18 ms 18 ms one.one.one.one [1.1.1.1]

Trace complete.

Edit: Fine again after reconnect. New IP is 155.93.247.8x
1-2 days in terms of disconnects means you're probably doing well on Octotel. :)

As for the routing, that is something we can/maybe sometimes can control.

Cloudflare will announce as they wish, and transit as they wish. It might change at the drop of a hat.

In the case of your trace, it seems like they took the traffic in CPT, but then routed it to Joburg (looking at the ping)

Unfortunately not much we can do about that. But will definately look into why your 155.93.247.x had a different profile. If it happens consistently...

If it is an issue we can control, we will help. But cloudflare routing is pretty dynamic.
 
Last edited:

TheRoDent

Cool Ideas Rep
Joined
Aug 6, 2003
Messages
5,695
What's the state of IPv6 currently? I had it working with WebSquad but had to disable it because there was something wrong causing issues.

I'm proud to say that you can ping any of our core and access routing infrastracture with V6, and I've been basically heads down on this for the last 3 weeks.

I took me one hour to enable our entire core with V6. (Around 200 routers) A script did the hard work.

I also have some cunning V6->V4 mapping but I'll leave that up to people that care about it. (and eventually github it).

Our website and hosting will be the last to get V6, as it's a little bit more complicated. Strangely enough, giving a consumer an IPV6 delegated address is logistically easier than managing tons of static AAAA records, and IP's for websites and servers, because delegation is more automatable via RADIUS.

So, it's actually easier to give consumers V6 than enabling every one of our CISP servers with V6. (But @Gimli) will probably use our V6 web accessibility as a measure. It's a good thing we're not a hosting farm, but a consumer ISP. Makes things easier.

In any case:
For example, one of our core routers is 154.0.0.1, but you can V6 ping it as 2c0f:f9a8::ffff:9a00:1, or 2c0f:f9a8::1

:ffff: being the canary for our V4 to V6 mapping, but we also have additional mappings. Nerds will get this. We are mapping the entire V4 space into our V6 space (because it's easy). In fact, you can ping any of our core/access infrastructure 154.0.0.X ip's using our 2c0f:f9a8::ffff "canary mapping" or simply 2c0f:f9a8::X

Enabling the core is technically a fairly simple part. Getting a core, and peering network V6 enabled is just some trundling.

But for most people V6 should be as normal as "ping google.com" (if we do it right)

Our next few weeks will be sorting out V6 pools for our PPPoE customers on an opt-in basis, as a trial.
Not expecting too much drama from that either.

I am a RADIUS expert, so I will keep your delegated V6 space fairly statically assigned, because that helps you as an end-user. There is some coding work to be done on that front from my side, but it's not ridiculously hard.

Our biggest problem wil be those people called Vumatel, especially their trenched network, because they control all of their network, and access via DHCP.

PPPoX based networks will be a no-brainer for enabling V6.

But internal technical support is where we will be focusing on. Especially focusing on CPE support for V6, and what it means to the end-user.

It's fine to have a V6 capable network, but without the people, and training behind it -- it's moot.

Caveat: We have by default, turned up IPV6 peering with all of our peers, where they support it, but it's likely that our V6 routing will be worse than our V4, for a while, since not not everyone peers on V6. But as always we will work on the best paths for our clients.

Which, getting back to @DrJohnZoidberg's point:

We don't want V6 to cause "issues", because the internet should just "work", right?

Our V6 deployment is a work in progress, and our goal is to make it our core, but also usable -- for moms, pops, consoles and gamers, and our support staff.

We have enough V4 space to keep people happy and service our customer base, but we want our V6 deployment to be equivalent, and better, without anyone knowing there's a difference between V4 and V6.

During our intitial deployment it will generally be opt-in, or opt-out. But we have to cater for all sides... Hence us taking our time...

Final Edit: [Jeez, ok, I ranted a bit] :cool:
 
Last edited:

Tinuva

The Magician
Joined
Feb 10, 2005
Messages
10,652
I'm proud to say that you can ping any of our core and access routing infrastracture with V6, and I've been basically heads down on this for the last 3 weeks.

I took me one hour to enable our entire core with V6. (Around 200 routers) A script did the hard work.

I also have some cunning V6->V4 mapping but I'll leave that up to people that care about it. (and eventually github it).

Our website and hosting will be the last to get V6, as it's a little bit more complicated. Strangely enough, giving a consumer an IPV6 delegated address is logistically easier than managing tons of static AAAA records, and IP's for websites and servers, because delegation is more automatable via RADIUS.

So, it's actually easier to give consumers V6 than enabling every one of our CISP servers with V6. (But @Gimli) will probably use our V6 web accessibility as a measure. It's a good thing we're not a hosting farm, but a consumer ISP. Makes things easier.

In any case:
For example, one of our core routers is 154.0.0.1, but you can V6 ping it as 2c0f:f9a8::ffff:9a00:1, or 2c0f:f9a8::1

:ffff: being the canary for our V4 to V6 mapping, but we also have additional mappings. Nerds will get this. We are mapping the entire V4 space into our V6 space (because it's easy). In fact, you can ping any of our core/access infrastructure 154.0.0.X ip's using our 2c0f:f9a8::ffff "canary mapping" or simply 2c0f:f9a8::X

Enabling the core is technically a fairly simple part. Getting a core, and peering network V6 enabled is just some trundling.

But for most people V6 should be as normal as "ping google.com" (if we do it right)

Our next few weeks will be sorting out V6 pools for our PPPoE customers on an opt-in basis, as a trial.
Not expecting too much drama from that either.

I am a RADIUS expert, so I will keep your delegated V6 space fairly statically assigned, because that helps you as an end-user. There is some coding work to be done on that front from my side, but it's not ridiculously hard.

Our biggest problem wil be those people called Vumatel, especially their trenched network, because they control all of their network, and access via DHCP.

PPPoX based networks will be a no-brainer for enabling V6.

But internal technical support is where we will be focusing on. Especially focusing on CPE support for V6, and what it means to the end-user.

It's fine to have a V6 capable network, but without the people, and training behind it -- it's moot.

Caveat: We have by default, turned up IPV6 peering with all of our peers, where they support it, but it's likely that our V6 routing will be worse than our V4, for a while, since not not everyone peers on V6. But as always we will work on the best paths for our clients.

Which, getting back to @DrJohnZoidberg's point:

We don't want V6 to cause "issues", because the internet should just "work", right?

Our V6 deployment is a work in progress, and our goal is to make it our core, but also usable -- for moms, pops, consoles and gamers, and our support staff.

We have enough V4 space to keep people happy and service our customer base, but we want our V6 deployment to be equivalent, and better, without anyone knowing there's a difference between V4 and V6.

During our intitial deployment it will generally be opt-in, or opt-out. But we have to cater for all sides... Hence us taking our time...

Final Edit: [Jeez, ok, I ranted a bit] :cool:
This is absolutely exciting news for technical people like me! Woohoo...finally!

Will there be an option of testing with a tunnel server (v6 over v4) for those interested and stuck on vumatel dhcp? (at least initial phase or something)

Everything else you doing sounds good. Maybe look into automating ip assignments to servers which looks like is the only thing you havent looked into automating when it comes to your websites/server infra (makes sense since you not focusing on providing hosting).
 

Gambit

Expert Member
Joined
Apr 26, 2005
Messages
1,578
Our biggest problem wil be those people called Vumatel, especially their trenched network, because they control all of their network, and access via DHCP.
Awesome update. Is Vumatel communicating anything to the ISPs about fixing this or is this unlikely to change in the foreseeable future?
 

Gambit

Expert Member
Joined
Apr 26, 2005
Messages
1,578
The update is "it should be supported soon" this was the feedback last week.
Even though there are no promises, that is good news that they have probably have assigned resources to it. I'm glad they didn't use the word "imminent" otherwise we will be stuck on IPv4.
 

PBCool

Cool Ideas
Company Rep
Joined
Jan 11, 2016
Messages
11,084
Even though there are no promises, that is good news that they have probably have assigned resources to it. I'm glad they didn't use the word "imminent" otherwise we will be stuck on IPv4.
It's quite strange because the trenched ecosystem is from the basis of a company from Sweden, so you would imagine being European based v6 would be a large requirement. I imagine it's just been on the backburner.
 

Pure_Acid

Senior Member
Joined
Nov 30, 2006
Messages
675
Anyone else have troubles in Sunningdale area in Table View, Cape Town on Frogfoot?

Below download has been like this since I woke up this morning on my 200mb line.

Ping: 4ms
Download: 0.75Mbps
Upload: 29.76Mbps
 

PBCool

Cool Ideas
Company Rep
Joined
Jan 11, 2016
Messages
11,084
Anyone else have troubles in Sunningdale area in Table View, Cape Town on Frogfoot?

Below download has been like this since I woke up this morning on my 200mb line.

Ping: 4ms
Download: 0.75Mbps
Upload: 29.76Mbps
If nothing on announcements please log it for me?
 
Top