iotaledger / iota / 29668768143
69%

DEFAULT BRANCH: develop
Ran
Jobs 1
Files 1126
Run time 2min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 68.816% (-0.003%) from 68.819%
29668768143

push

github

web-flow 
feat(starfish-config): reduce default soft_leader_timeout to 5 ms (#12298)

# Description of change

Reduce the default StarfishSpeed `soft_leader_timeout` from 100 ms to 5
ms. Local operational parameter only — no protocol version change;
per-authority inconsistency is tolerated by design.

Rationale, from a 10-run sweep of the timer (0-100 ms) on a 20-validator
network with the geodistributed latency matrix (#12280):

- Commit latency and commits/s are insensitive to the timer across the
whole range (tx commit p50 254-263 ms, p95 740-745 ms, ~2000 commits/s
everywhere).
- Smaller values let high-latency validators propose without waiting on
strong-vote propagation: at the low end their own-block production
reaches full parity with the rest of the committee (16.2 blocks/s across
all 20 validators at 0 ms), and their leader votes leave earlier — which
is what reputation scoring rewards, since a vote only earns credit while
other validators still strong-link it.
- Proposing on soft timeout never skips the leader link
(`Core::try_new_block` requires the parent-round leader header for
non-forced proposals), so no votes are forfeited by proposing earlier.
- 5 ms rather than 0: the strong-vote readiness check stays armed for
votes arriving together with the parent-round quorum, instead of being
bypassed the instant a round opens. Behaviorally near-identical on
geodistributed topologies (SoftTimeout proposal share 16.3% vs 16.9%).

Full sweep data and discussion: #12280.

## Links to any relevant issues

#12280

## How the change has been tested

- [x] Basic tests (linting, compilation, formatting, unit/integration
tests)
- [x] Patch-specific tests (correctness, functionality coverage)
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [x] I have checked that new and existing unit tests pass locally with
my changes

`starfish-config` (12) and `starfish-core --lib` (336) tests pass with
the new default. The 5 ms value itself w... (continued)

167180 of 242937 relevant lines covered (68.82%)

485524.67 hits per line

Coverage Regressions

Lines Coverage ∆ File
16
82.81
 -0.48% crates/iota-core/src/authority.rs
12
93.83
 -0.44% crates/starfish/core/src/authority_service.rs
7
91.55
 -0.32% crates/starfish/core/src/header_synchronizer.rs
3
93.47
 3.44% crates/iota-core/src/connection_monitor.rs
2
85.64
 -0.49% crates/iota-core/src/storage.rs
2
89.03
 -0.27% crates/iota-network/src/discovery/mod.rs
2
75.18
 0.29% crates/iota-network/src/randomness/mod.rs
1
91.04
 0.04% crates/iota-core/src/authority/authority_per_epoch_store.rs
1
84.73
 -0.49% crates/iota-core/src/global_state_hasher.rs
1
84.96
 0.27% crates/iota-json-rpc/src/read_api.rs
Jobs
ID Job ID Ran Files Coverage
1 nextest+simtest - 29668768143.1 1126
68.82
 GitHub Action Run
Source Files on build 29668768143