Skip to content
  • Automatic Updates
  • Firmware Signing
  • Monitoring Concept
  • Administration Concept
  • Users Agreement
  • Internet Access realization
  • IP Management solution to have no conflicts across all clusters
  • Mesh Security
  • IPv6 TTL / Hop Limit, when will this become a problem in huge mesh networks?

Foster Nodes in small clusters

Especially small cluster maybe don't want to run any Foster Nodes whatsoever. What to do about the most concerning Foster Services like - DNS - Update Service - Monitoring Maybe small clusters are being offered those services through the Internet which of course will make an Internet uplink mandatory.

IPv6 TTL

In larger Mesh networks the 8 Bit TTL / Hop Limit might be a show stopper. Thus we need to address this issue. One option would be to use a Layer 2 Dynamic Routing Solution like batman-adv. This way all nodes would be able to see each other like if they all were connected to a switch. In order to be able for clients to reach each other via IP it seems necessary to also use a Layer 3 Routing Algo on top.

Node Owner Admin Access

The topic is up for discussion. Either A) give the owner admin access to their node B) prohibit admin access to the owner's node Each of those two options has its ups and downs.

Let's assume the owner will not gain admin permission to their node. Does this imply they will have to install their very own generic firewall in order to manage settings regarding firewall / NAT and other stuff. So it seems mandatory that nodes should be setup in a way that owners are able to connect their very own router. On the other hand when an owner does not care about admin access at all and just wants to get basic Internet connectivity it seems like a valid option to configure a port on the node with a primary VLAN id that offers just basic Internet access.

Automatic Updates

  • possbile with Ansible?
  • how does Freifunk do it?
  • how does OpenWisp do it?

When we suppose that pplznet will not generate one firmware per cluster but one firmware per node it will be much easier to do upgrades. Using imagebuilder packages and configuration will directly burned into the image. An update would mean just to flash the new, updated image. Question left is how to automate the process. We will need to collect experience with this technique.

Firmware signing

  • how does Freifunk do it?

Monitoring

  • metrics
  • crosslink monitoring
  • https://datatracker.ietf.org/doc/html/rfc9046
  • Notifications
  • Zabbix / Nagios / Icinga ??? Will each cluster need its own Monitoring instance?

Administration Concept

  • use Ansible to orchestrate all nodes?
  • would need consent from node owners.

Internet Access

Paid pplsznet users receive high speed Internet access. How can this be realized ideally? We need authentication mechanisms but at the same time performance should not suffer too much.

  • Wireguard VPN Gateway?
  • Splash Page?
  • AAA (Authentication / Authorization / Accounting)
  • Radius?
  • Kerberos?
  • LDAP?
  • ...

Mesh Security

https://datatracker.ietf.org/doc/html/rfc8967 https://datatracker.ietf.org/doc/html/rfc8968