DAOS-18825 pool: include excluded ranks as DOWNOUT at creation time - #18994
Open
wangshilong wants to merge 1 commit into
Open
DAOS-18825 pool: include excluded ranks as DOWNOUT at creation time#18994wangshilong wants to merge 1 commit into
wangshilong wants to merge 1 commit into
Conversation
|
Ticket title is 'Imbalanced object distribution for G32 oclass in pool with asymmetric fault domain topology' |
wangshilong
force-pushed
the
shilongw/DAOS-18825-2.8
branch
from
September 3, 2026 08:47
8a2e6c6 to
4987873
Compare
DAOS placement uses consistent hashing to distribute data across nodes and fault domains. In theory, each node/domain should have the same probability of being selected for data placement, which helps ensure balanced data distribution. However, pool create currently filters out excluded ranks and only creates VOS pools on alive ranks. If a node has only one alive rank, this can cause data imbalance. To address this, pool create now sends DOWNOUT ranks to the PS leader. VOS pools are still created only on alive ranks, but DOWNOUT ranks are included in the pool map. During layout calculation, placement remaps data from DOWNOUT ranks onto alive ranks, reducing imbalance. After reintegration, data can be moved back to the reintegrated ranks. Features: rebuild pool Signed-off-by: Wang Shilong <shilong.wang@hpe.com> Signed-off-by: Xuezhao Liu <xuezhao.liu@hpe.com> Co-authored-by: Xuezhao Liu <xuezhao.liu@hpe.com>
wangshilong
force-pushed
the
shilongw/DAOS-18825-2.8
branch
from
September 7, 2026 02:20
4987873 to
049287e
Compare
wangshilong
marked this pull request as ready for review
September 7, 2026 02:31
knard38
approved these changes
Sep 7, 2026
tanabarr
approved these changes
Sep 7, 2026
Collaborator
|
Test stage Functional Hardware Medium MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18994/3/execution/node/1672/log |
liuxuezhao
approved these changes
Sep 8, 2026
Collaborator
|
Test stage Functional Hardware Medium MD on SSD completed with status FAILURE. https://jenkins-3.daos.hpc.amslabs.hpecorp.net//job/daos-stack/job/daos/view/change-requests/job/PR-18994/4/execution/node/610/log |
daltonbohning
left a comment
Contributor
There was a problem hiding this comment.
ftest changes LGTM (they match what is in master)
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
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.
DAOS placement uses consistent hashing to distribute data across nodes and fault domains. In theory, each node/domain should have the same probability of being selected for data placement, which helps ensure balanced data distribution.
However, pool create currently filters out excluded ranks and only creates VOS pools on alive ranks. If a node has only one alive rank, this can cause data imbalance. To address this, pool create now sends DOWNOUT ranks to the PS leader. VOS pools are still created only on alive ranks, but DOWNOUT ranks are included in the pool map. During layout calculation, placement remaps data from DOWNOUT ranks onto alive ranks, reducing imbalance. After reintegration, data can be moved back to the reintegrated ranks.
Features: rebuild pool
Steps for the author:
After all prior steps are complete: