Zero OS does not barf if node is only connected via IPV4

[andhartl]

Farmers like GEC connect nodes on ipv4 addresses only. they are listed as node in the explorer but can not be connected to for capacity usage. can we implement a mechanism which doe snot allow ipv4 only connection in combination listing? thx

[zaibon]

This is becoming a more present issue now.
There seems to be a lot of nodes that are booted but don't have a proper network configuration:
http://10.4.0.250:3000/explore?orgId=1&left=%5B%22now-1h%22,%22now%22,%22Loki%22,%7B%22expr%22:%22%7Blevel%3D%5C%22fatal%5C%22%7D%22%7D,%7B%22mode%22:%22Logs%22%7D,%7B%22ui%22:%5Btrue,true,true,%22none%22%5D%7D%5D

To solve this we can send a mail to the farmers. I propose we create a script that query all the nodes that report this kind of error, we extract the node ID from the result, then send a mail to all farmer concerned to invite them to review their IPV6 network farm configuration.

[muhamadazmy]

@zaibon Can we check this only by revising nodes public_config and then see if they have ipv6 or not? Also public_config = null is still allowed/possible, no?

[zaibon]

@zaibon Can we check this only by revising nodes public_config and then see if they have ipv6 or not? Also public_config = null is still allowed/possible, no?

No this is not how it works. public_config is an optional configuration. Most of the node won't have this. I don't think we have that information in the explorer at the moment.

CC @delandtj

[zaibon]

can we implement a mechanism which doe snot allow ipv4 only connection in combination listing? thx

instead I prefer we do the following. The node is still registered in the explorer. So the farmer can have some information about the network interfaces, capacity and he can earns its farming token.
But in addition to this, we need a way to show that the node network is not right.

An extra field in the node object ?
Maybe we could have some kind of health check for each daemon. And report this health check to the explorer ?

@DylanVerstraete @muhamadazmy

[DylanVerstraete]

sounds like a good idea, this will force the farmer not as much as sending him an email about some missing configuration. But then again, will the farmer check the explorer and notice these health checks?

[muhamadazmy]

@zaibon Don't we already report all interfaces of the node to the explorer ? Can't we check this and show a warning if no interface has a valid ipv6 ?
Implementing a health checks for the daemon is more generic and better idea but is way much work.

[zaibon]

Don't we already report all interfaces of the node to the explorer

We do it seems we do not report the info for the ndmz network namespace and interface, which is where we can detect if the node is all good or not.
I guess the code reporting the interfaces needs to be adapted to also include it

[DylanVerstraete]

Whilst looking for reporting the ndmz network namespace I noticed that the macaddress for all the interfaces is not properly reported.
Also we seem to report the ndmz bridge:

"ifaces": [
{
"name": "lo",
"addrs": [
"127.0.0.1/8",
"::1/128"
],
"gateway": null,
"macaddress": ""
},
{
"name": "eth0",
"addrs": [
"fe80::5645:46ff:fef6:261/64"
],
"gateway": null,
"macaddress": ""
},
{
"name": "zos",
"addrs": [
"192.168.123.22/24",
"fd74:cc1c:6079:0:5645:46ff:fef6:261/64",
"fe80::ac15:adff:fe7b:b3ae/64"
],
"gateway": [
"192.168.123.1"
],
"macaddress": ""
},
{
"name": "br-ndmz",
"addrs": null,
"gateway": null,
"macaddress": ""
}
],

[zaibon]

We need do not care about the IP address of the ndmz bridge. We want the IP of the interface in the ndmz network namespace.