subsystem-bench: adjust test config to Kusama #3583
Conversation
AndreiEres
added
R0-silent
We could've used these values assuming 90 and 99 percentiles min_pov_size = 64 // 108 KiB, 90%
max_pov_size = 2024 // 1.78 MiB, 99% |
...t/node/network/availability-distribution/tests/availability-distribution-regression-bench.rs
Outdated
Show resolved
Hide resolved
| fn default_pov_size() -> usize { | ||
| 5120 | ||
| 2 * 1024 |
There was a problem hiding this comment.
Let's use max value here such that any degradation has more impact.
| fn default_backing_group_size() -> usize { | ||
| 5 | ||
| 3 |
There was a problem hiding this comment.
This has to be kept in sync with the actual value from chain. Let's keep it 5 here and override in specific benchmarks.
| 300 | ||
| } | ||
|
|
||
| // Based on number of kusama validators producing blocks |
There was a problem hiding this comment.
This is not related to the validators producing blocks. It's the total number of cores: cores assigned to specific paras plus on-demand (chain config value)
There was a problem hiding this comment.
But we don't want to fetch all cores. We'd want to represent a single node effort. Roughly this number has to be set to n_cores * needed_approvals / n_validators.
There was a problem hiding this comment.
n_cores * needed_approvals / n_validators. only so for availability read, for write, we need to cover all cores as said above.
Fixes paritytech#3528 ```rust latency: mean_latency_ms = 30 // common sense std_dev = 2.0 // common sense n_validators = 300 // max number of validators, from chain config n_cores = 60 // 300/5 max_validators_per_core = 5 // default min_pov_size = 5120 // max max_pov_size = 5120 // max peer_bandwidth = 44040192 // from the Parity's kusama validators bandwidth = 44040192 // from the Parity's kusama validators connectivity = 90 // we need to be connected to 90-95% of peers ```
Fixes #3528