[go: up one dir, main page]

Tezt_cloud/DAL: Use produce_regularly to produce from the DAL node instead of relying on the orchestrator

Guillaume Genestierrequested to merge
GuillaumeG@producer_regularly into master

What

Add an option --produce-regularly in tezt-cloud to ask the DAL node to publish data regularly "autonomously".

Why

Because, when tezt-cloud experiments involve a lot of machines, the orchestrator is overwhelmed and miss a lot of publication orders. This aims at lightning the burden on the orchestrator by letting the producer DAL nodes deal of publications by themselves.

How

By using the feature added in the DAL node by !19975 (merged)

Manually testing the MR

One can do a very light test with a configuration file like:

{
  "localhost": true,
  "website": true,
  "website_port": 8080,
  "dockerfile_alias": "dal",
  "scenario_specific":
    [ "DAL",
      { "stake": [1],
        "producers": 1,
        "produce_regularly": true
      }
    ]
}

And then launch the experiment with:

make && dune exec tezt/tests/cloud/main.exe DAL -- -v --config-file config.json --tezt-cloud YOUR_ALIAS --log-file /tmp/$(date "+%Hh%M").log

Merge request reports