lodestar
lodestar copied to clipboard
ChainSafe
Sync committee participation sometimes fell down 50% for few slots in the epoch
Describe the bug
Observed randomly among different runs of simulation tests that sync committee participation fell down the 50% for few slots in the epoch. Which end up having less than 90% average for the epoch with the epoch length of 8 slots for minimal preset.
Slots which are not listed in table below have participation values 1.
| Slot | node-1-beacon-lodestar | node-2-beacon-lighthouse |
|---|---|---|
| 1 | - | - |
| 2 | - | - |
| 3 | - | - |
| 13 | - | - |
| 14 | - | - |
| 15 | - | - |
| 16 | 0 | 0 |
| 17 | 0.40625 | 0.40625 |
| 18 | 1 | 1 |
| 19 | 1 | 1 |
| 20 | 1 | 1 |
| 39 | 1 | 1 |
| 40 | 1 | 1 |
| 41 | 0.40625 | 0.40625 |
| 42 | 1 | 1 |
| 43 | 0.40625 | 0.40625 |
| 44 | 1 | 1 |
| 45 | 1 | 1 |
| 46 | 1 | 1 |
| 47 | 1 | 1 |
Expected behavior
There should be 100% participation for all slots except the first slot of the altair fork.
Steps to reproduce
Run sim tests in multiple times.
Additional context
Check the multi-client logs for the following sim-run attached.
Operating system
Linux
Lodestar version or commit hash
unstable
For reference this issue is not new and I observed this behavior many times earlier as well. But in multi_fork test because we have 5 nodes, so low participation for 1 or 2 nodes in few slot does not reduce the epoch participation below 90%. In multi-client test we have only 2 nodes, so why this impact a lot in this test.
We should revert this change when this issue is fixed. https://github.com/ChainSafe/lodestar/blob/5047410270ed7d74f686d6fab3e9d0217a6bcf6a/packages/cli/test/utils/simulation/assertions/defaults/syncCommitteeParticipationAssertion.ts#L6-L8