This was partially only a logging issue, that logged this in the collator protocol subsystem.
The issue of fluctuating block times is a known issue and is planned to be fixed with the v0.9.19 release (at the time of writing).
drahnr
- 386
- 2
- 7