Stability area meta + timeline

[zelig]

Tooling for network tests

User-experience benchmarks and metrics

• upload-download success rate - Baseline benchmark of upload and download success rate #1033
• upload speed for different file sizes Upload speed for different file sizes #1091
  • hash response (split to chunks complete)
  • wait for local storage
• download speed for different file sizes Download speed for different file sizes #1092
  • lowest latency
  • highest
  • average/median

Need more work to define

• failure rate of accessing existing content long term
• availability delay for different file sizes
  • earliest node can retrieve
  • all nodes can retrieve

Network testing

• GlobalStore - shall we complete this, and add it as part of the Swarm app/deployment charts?
• Chunk Explorer - shall we implement this tool - basically the ability to inspect chunks on given nodes?
• Implement a tool to check if there is a path from one node to another based on syncing streams (server/client)?

Simulation framework

• swarm: snapshot load improvement ethereum/go-ethereum#18220
• swarm/network, swarm/pss, p2p/simulations: adapt test kademlia params #1022
• remote (docker) adapters monitoring - Remote docker adapter for simulations #991
• dropout models
• network contention models

Mocking tools

• reinstate mock db via global store
• allow chunkless syncing - Mock chunk sending for more robust simulations #992

k8s

• Logging for k8s containers (maybe ELK / EFK - https://akomljen.com/get-kubernetes-logs-with-efk-stack-in-5-minutes/ ?)~
• Container logs rotation PR#58~
• Ingress for services~
  • for developers to individual Swarm pods~
  • for developers to Logging and Monitoring and Kubernetes Dashboard~
  • for users to Swarm, load balanced across all Swarm pods~
  • for monitoring, logging, and other auxiliary services~
• SSL for all ingress services~
• Authentication/Authorization for developers (AWS IAM access/secret keys, with no access to AWS services or to AWS console) to~
  • k8s / kubectl - it would be good if devs have access to this, so that they can port-forward and docker exec towards specific swarm containers.~
  • Monitoring~
  • Logging~
• Auto-scaling of staging cluster
• Documenting the setup
  • Terraform for the initial AWS EKS setup~
    • Variables, such as different launch groups, and instance types, and why we need them~
  • Kubernetes Dashboard~
  • Prometheus Monitoring
  • Logging
  • Geth application and how to deploy a new version
  • Swarm application and how to deploy a new version~
• AWS Usage/Costs dashboard~
• Monitoring of PV EBS drives - the dashboards that come with the current Monitoring stack don't have this.~
• Hardening for:
  • Logging setup.
  • Metrics setup.
  • All Roles and Service Accounts created by external helm applications we use.
  • OS patching?
• Upgrade strategy for k8s

[petyaracz]

upload speed for different file sizes

what's the diff between hash response and wait for local storage ?

[zelig]

@petyaracz disk storage is asyncronous, it does not finish strictly by the time the chunker (and the API) returns the swarm hash.