HDFS-17936. Detect slow disks separately by storage type#8558
Open
lfxy wants to merge 1 commit into
Open
Conversation
|
🎊 +1 overall
This message was automatically generated. |
lfxy
force-pushed
the
featrue/HDFS-17936
branch
2 times, most recently
from
2972727 to
f3d397f
Compare
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The current slow disk detection does not differentiate between storage types. When a DataNode has 1 SSD and 10 HDDs with dfs.datanode.max.slowdisks.to.exclude=2 and
dfs.datanode.min.outlier.detection.disks=5, if the SSD is under heavy read/write load, the single SSD may be identified as a slow disk. Client tasks writing SSD data to that DataNode will then fail to select a volume and report: No more available volumes.
public V chooseVolume(List<V> volumes, long replicaSize, String storageId) throws IOException { if (volumes.size() < 1) { throw new DiskOutOfSpaceException("No more available volumes"); } ... }Therefore, slow disk detection should be performed separately for each storage type. If the number of disks of a given storage type is less than dfs.datanode.min.outlier.detection.disks, detection for that storage type is skipped entirely.