HDDS-15093. Make RackScatter intra-rack placement capacity-aware#10648
Open
chihsuan wants to merge 1 commit into
Open
HDDS-15093. Make RackScatter intra-rack placement capacity-aware#10648chihsuan wants to merge 1 commit into
chihsuan wants to merge 1 commit into
Conversation
chihsuan
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What changes were proposed in this pull request?
The
RackScatterEC placement policy scatters replicas across racks, but inside a rack, it picks a datanode uniformly at random with only a binary "has enough space" check. In a heterogeneous rack (e.g., a 100TB and a 400TB datanode), both are equally likely to be chosen, so the smaller node fills up first.Fix. Make the intra-rack choice capacity-aware, reusing the power-of-two approach already used by
SCMContainerPlacementCapacity: pick two distinct candidate nodes in the rack and keep the one with lower space utilization. This biases placement toward emptier nodes without herding every write onto a single node.Whether a node has enough space is still gated by the existing
isValidNodecheck. This only changes which valid node is preferred.What is the link to the Apache JIRA
https://issues.apache.org/jira/browse/HDDS-15093
How was this patch tested?
TestSCMContainerPlacementRackScatter#chooseNodeWithinRackPrefersLessUtilized: a single rack with an emptier and a fuller datanode; over 1000 placements the emptier node is chosen significantly more often.TestSCMContainerPlacementRackScattersuite still passes (100 tests).checkstyle.shclean on the module.