[EPIC] Telemetry tracking issue

[liamsi]

Telemetry

We've started the telemetry integration process as outlined in #901, and this issue gathers all the sub-issues and/or PRs related to the integration/implementation process.

Base(d)

Scaffolding for further integration

• [ ] #663
• [ ] https://github.com/celestiaorg/celestia-node/pull/907
• [x] https://github.com/celestiaorg/celestia-node/issues/934
• [ ] #935
• [ ] #936
• [ ] #937
• [ ] https://github.com/celestiaorg/celestia-node/issues/1003

Metrics Integration Spots

Actual code points for us to monitor/measure/observe/etc

• [ ] https://github.com/celestiaorg/celestia-node/issues/639
• [ ] https://github.com/celestiaorg/celestia-node/issues/98
• [x] https://github.com/celestiaorg/celestia-node/issues/999
• [x] https://github.com/celestiaorg/celestia-node/issues/1040
• [ ] Uptime metric

Tracing Integration Spots

• [ ] ipld: consider tracing for GetShares
• [ ] share: simple trace for Light (besides Full) Availability

[tac0turtle]

this would be awesome!!

[Wondertan]

Proposing https://github.com/VictoriaMetrics/VictoriaMetrics

[cemozerr]

I'll try chew on this. It seems like geth's metrics package is a fork of the go-metrics package with prometheus support added. I'll try and integrate that to celestia-node.

LETS GET THAT MONITORING SETUP ONLINE

[liamsi]

Key metrics of the top off my head:

• measure sampling time
• measure BlockSync time (per block and in total from height h, where h could be 0/genesis) for full nodes
• header sync time
• ....

[renaynay]

@liamsi sync is different from DAS. sync is only header sync rather than the actual block reconstruction. That happens during DAS