Run without chainNodes/privValAddr

[activenodes]

@agouin thank you for this explanation https://github.com/strangelove-ventures/horcrux/issues/100

What do you think of this "hybrid" setup ?

2 Datacenter (with in each: 1 sentry and 1 signer)
1 Spare Datacenter (with 1 signer and without sentry)

DC01
Sentry01 <-> Signer01 <->|to S. 2+3

DC02
Sentry02 <-> Signer02 <->|to S. 1+3

DC03
             Signer03 <->|to S. 1+2

this way each sentry is connected to its own local signer (latency 0) and there is an additional signer reachable p2p by the other signers when a DC is down.

I think it's the minimum configuration, but with high performance and full redundancy.

Signers connected to all sentries may cause connections cross like Sentry01<->Signer02 and Sentry02<->Signer01. Right?

So in the Signer03 we need to run without "chainNodes/privValAddr" but the result is:

Error: need to have chainNodes configured for priv-val connection

[activenodes]

Otherwise we could make this dynamic:

horcrux/signer/remote_signer.go

Line 21 in 03f565d

const connRetrySec = 2

and insert a dummy IP.

Example:

chainNodes:
- privValAddr: tcp://127.0.0.1:1234
  connRetrySec: 99999

[agouin]

This would be a nice feature. We will get this in the next release

[activenodes]

This would be a nice feature. We will get this in the next release

Thank you!
I'm migrating from some TMKMS clusters to Horcrux, testing it in every way... great development

[activenodes]

@agouin please remember this 🙏

Note. the original idea was to run a cosigner without "chainNodes/privValAddr", without changing connRetrySec in a variable

[agouin]

This is now possible in v3.2.0