3

What are some of the methods people are using to make sure a baking node remains up and running 24/7? I see all these baking services with "enterprise server solutions", whatever that means. However, when node falls over, it still needs someone to fix/restart etc. What if the owner of the node is not currently available or perhaps had to travel to a different country. Would SSH remote access be an option? What are some of the ways people reduce risks here? Thanks

1

We are or hire system architects who design a HA (high availability) infrastructure for our bakers πŸ˜‰ You can see some discussions here. SSH is definately one tool in the toolbelt here, but if you want a serious HA architecture you need to find skilled people to build and operate it πŸ‘

2
  • Thanks! SSH because it can be used for remotely accessing the machine for troubleshooting? – Argonau7 Apr 2 '19 at 12:18
  • Yes πŸ˜„πŸ‘ If you open ssh to the internet, I would recommend only allow login using keys (and not passwords). Preferrably have your nodes on a VPN and allow remote access only via the VPN. Feel free to each out on a PM on twitter @asbjornenge or the Obsidian slack channel if you need assistance setting up your infra πŸ˜‰πŸ‘ – asbjornenge Apr 3 '19 at 12:35
1

A lot of us bakers old timers depend on the work done by etomknudsen called "Tezos Node CPR".

Main repo has systemd scripts to let systemd manage and log the tezos binaries. Systemd does a great job keeping services up as it's battle tested in enterprise level services, it handles process dependencies and restarting of services on errors.

The Tezos Node CPR is an optional service that monitors and restarts the above Tezos services as it detects problems such as:

  • Node hasn't gotten new blocks for N seconds
  • Network disconnected
  • RPC on the node is down

The common problem it solves is IP changes due to dynamic ip. The node itself doesn't have logic to "reset" itself into a working state once the node becomes an invalid peer. The only way to correct the problem is by restarting the process.

That said, Tezos node CPR doesn't replace an actual administrator as baking may require an update to the binaries at any moment due to hotfixes and features upgrades. Baking is not a complete passive endeavor.

Your Answer

By clicking β€œPost Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.