ClientDelay seems to be incorrectly processed

[bergerch]

In orchestrator.js a template should be created where the start_time of client processes is set to clientDelay as described in the experimental description file.

shadowTemplate = await createShadowHostConfig(
      shadowTemplate,
      hosts,
      e[experimentId].misc.clientDelay
);

but then createShadowHostConfigseems to ignore the passed clientDelay param:

async function createShadowHostConfig(
  shadowTemplate,
  replicas,
  clientDelay
) {
  for (let i = 0; i < replicas.length; i++) {
    shadowTemplate = yg.makeHost(
      shadowTemplate,
      replicas[i].name,
      replicas[i].ip,
      i,
      replicas[i].procs
    );
  }
  return shadowTemplate;
}

the resulting conf file always has a start_time: 0 even if clientDelay is specified. This seems to bug experiments with BFT-SMaRt if clients send requests before replicas are ready.

Edit: Not sure if this bug was fixed in the meantime, I am a little behind the most recent main branch. (Feel free to close if it is fixed now)

[bentoumia]

@bergerch good catch :D this was fixed, use exp.client.startTime (optional; default: 0) instead see:

tool/examples/serial22/bftsmart-AWS.yaml

Line 32 in b3d912f

startTime: 30 s

[bentoumia]

Closing as its fixed.