Contributors: @vect0r @gabriel.moise @GuillaumeGen @mattiasdrp @4dam

Context & Goal

Following the bandwidth assessment of the DAL layer, the team identified a set of maintainability, usability, and performance issues in the tezt-cloud platform — the main testing tool for experimentation. This milestone aims to enhance the robustness, developer experience, and reusability of tezt-cloud, with a view toward upcoming projects and future experimentation cycles that will rely on this infrastructure.

Reference: Tezt-cloud: a unified test framework

Objectives & Tasks

tezt-cloud Improvements

• Scenario Clean-up and Refactoring (@gabriel.moise)
  • Deduplicate logic between DAL and L1 scenarios
    • Use insights from Anne-Laure closed MR
    • Unify snapshot utilities - !18729 (merged)
    • Refactor agent_kind, yes_crypto and node - !18767 (merged)
    • Refactor dal node - !18768 (merged)
    • Refactor etherlink - !18769 (merged)
    • Refactor baker - !18785 (merged)
    • Refactor metrics and monitoring - !18788 (merged)
    • Refactor node_helpers utilities - !18854 (merged)
  • Improve documentation
    • Improve top-level README
      • Explain project layout, basic command usage and development flow
      • Add documentation for each scenario, including scenario purpose, useful CLI arguments, potential issues/TODOs
      • Document internal helper functions
    • Documentation for scenarios: !18865 (merged)
• Rich Accounts via Yes-wallet (@vect0r)
  • Enable bootstrapping experiments with high-balance accounts !18715 (merged)
• Configuration & Reusability (@GuillaumeGen)
  • Potential reference : !16143 (closed)
  • Introduce configuration file support
    • Replace verbose CLIs with structured, reproducible configs
    • Suggested fields: machines, ENV vars, Octez version, protocol, test variables etc.
  • Allow starting from a stored snapshot
    • From GCP bucket or local via ENV var
  • Dockerfile improvements
    • For instance, base -> dal -> dal+snapshots/rollups
  • Investigate, and fix, the --localhost memory over consumption
  • Enable snapshot import via GCP/public URLs
  • Introduce --richest-accounts-over 10, 1000 command line option for the yes-wallet !18715 (merged)
• Automated Logging & Monitoring (@vect0r)
  • Make sure Teztale is available and functioning properly
  • Make sure Netdata is available and functioning properly
  • Retrieve daily logs from L1 nodes, bakers, DAL nodes etc. locally !18828 (merged)
    • Followup cleanup !18888 (merged)
  • Upload daily logs to GCP bucket
  • Collect and expose:
    • Prometheus/Teztale DB stats
    • Profiling data
    • OpenTelemetry traces
• macOS Compatibility (@4dam)
  • Reference: %tezt-cloud macOS compatibility
  • Gather a list of current issues with running the tezt-cloud experiments on macOS distribution
  • Prepare a plan on how to tackle them
• Minor issues/bugfixes @vect0r
  • Fix the "we lack of information" by using delegate keys !18716 (merged)
  • Remove CLI argument --force-history-mode-switch !18724 (closed)
  • Fail gracefully if NUM_OF_PRODUCERS + NUM_OF_DAL_SLOTS > 32
  • Factor out History Mode 79 logic !18725 (merged)

Handle GCP deprecation policy

• Migrate to "migrating to container deployment startup scripts"
  • See https://gitlab.com/tezos/tezos/-/blob/master/tezt/lib_cloud/terraform/vm/main.tf#L121
  • !16117 (closed)
  • !17141

DAL scenario improvements

• Profiler in DAL nodes
  • Profile DAL node and GossipSub (@gabriel.moise)
  • Retrieving data improvements
• OpenTelemetry DAL operations
  • Gather more data (e.g. Message P2P timing)

RBT scenario needs

• OpenTelemetry consensus operations
  • Add filtering from all operations

Store backends needs (@mattiasdrp)

• Add config-file toggle to switch store backends
  • Moved from CLI to configuration file
• Support non homogenous store backends
• Support specific Tezedge snapshots
• Add scenarios for the project needs