monitoring page for Market

[tworec]

Continuation of #506

Conclusions (repeated from gdoc)

Required

• Monitor would request for yagna REST endpoints
  • It needs to expose statistics @ new endpoint (REST)
  • No local persistence of statistics
• Monitor would request only known node (nodes)
• We need two services
• Metric collector service (eg. prometheus + grafana; timeseries; quantitative)
• Dashboard / charts configuration
• Yagna daemon GUI (live; only local data; detailed; custom frontend)

Optional

• Broader network info will be retrieved using own nodes (via Offers)
• Provider would be also monitored

Comments

• We are not expecting new provider incarnations
• We do not care to other (hypothetical) Provider compatibility

PoC functional requirements:

• enough is to have a snapshot of current network status (history not needed)
• UI: CLI is enough, (Grafana GUI is nice-to-have/optional)
• single Metrics:
• agreements cnt

MVP functional reqs:

• all above and
• UI depends on results of PoC
• Metrics
  • nodes cnt (per status: running, shut down)
  • offers cnt
  • demands cnt
  • agreements cnt (per status)
  • activities cnt
  • invoices cnt

Version with history:

• central active backend storage (Prometeus)
  • active means it sends requests to nodes
• simplistic UI: GUI or CLI (which is simpler) eg. Grafana

Originally posted by @tworec in #506 (comment)