Revert AMM ReduceReplicas and parallel AMMs updates #5335
Merged
+145
−181
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A group recently reported offline that they were running into deadlocked clusters with the current
distributedmainbranch (xref dask/community#182 (comment)). After runninggit bisectthey were able to narrow down #5297 as the place where things began to lock up.As #5315 and #5297 both contain updates to our active memory management system, which is still under active development and disabled by default, this PR proposes we revert those two PRs prior to the 2021.09.1 release. This will help us avoid having users impacted by the deadlock issue and give @fjetter @crusaderky and me more time to determine what exactly in #5315 and #5297 is problematic.
cc @crusaderky @fjetter @jakirkham