iotaledger / iota / 27079501074

Build # 27079501074

Build Type

push

github

Committed by web-flow

Commit Message

feat(migration-test): deterministic realistic latency model with heavy-tail validators (#11628)

# Description of change

Dev-tools only (`dev-tools/iota-private-network/experiments`).

- **Realistic latency model** — `network-benchmark.sh` replaces the
symmetric 10×10
RTT table with a deterministic, directed, role-based matrix (hub / band
/
relay follower / heavy tail, repeating every 10 validators) for any
validator
count, exercising consensus paths a benign matrix never hits
(header-relay
completion, round skipping, block-rate spread). `-D` dumps the effective
  matrix, `-L` overrides it. Validated live for n = 10–24 and 30.
- **Latency-injection reliability** — a transient `tc qdisc show`
failure no longer
wipes applied netem; the watcher heals partial wipes and survives
transient
errors; initial application no longer races on the xtables lock;
lockfiles
  moved out of `/var/lock`.
- **Python runners replace the bash suites** — `run-all-benchmark.sh` /
`run-all-fuzz.sh` are removed in favor of `run-benchmark.py` /
`run-fuzz.py`
on a shared `experiment_common.py` (dynamic compose generation for any
n,
upfront sudo caching, a cross-runner single-run lock, the stress load
image
resolved up front — pulled, else built from a `network-benchmark` clone
—
with load started before the block-production measurement so it runs
under
  load, and Grafana force-recreated on the correct docker network);
  `run-migration-test.py` is refactored onto the same base.
- **Migration-test modes** — `--mode {simple,advanced}` (default simple:
rolling
upgrade with a pre/post block-production comparison; advanced keeps the
full
  restart-torture schedule). Defaults: testnet image, 10 validators.

## Links to any relevant issues

Fixes #11664
Fixes #11738
Fixes #11739
Fixes #11740
Fixes #11741

## How the change has been tested

- [x] Basic tests (linting, compilation, formatting, unit/integration
tests)
- [x] Patch-specific tests (correctness, functionality coverage)
- [ ] ... (continued)

Coverage Stats

152828 of 227763 relevant lines covered (67.1%)

468935.41 hits per line

Source File

Press 'n' to go to next uncovered line, 'b' for previous

Source Not Available