There are a few issues which are possibly related to the new GC pacer (e.g. and #50705 and #50956).

Furthermore, the implementation currently in 1.10-beta2 and master seems to deviate from what has been originally implemented in v8 (see discussion in #50909).

Due to these reasons, we discussed in triage that it might be better to de-risk the release by reverting the MemBalancer implementation, but keep the per-page memory accounting introduced in that PR.

CC: @NHDaly, @gbaraldi, @vchuravy