HDFS-17299. Adding rack failure tolerance when creating a new file #6513
Conversation
This file contains 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
|
💔 -1 overall
This message was automatically generated. |
added 3 commits
January 31, 2024 16:50
… Fixing test failures
… Fixing test failures
… Fixing test failures
|
💔 -1 overall
This message was automatically generated. |
added 2 commits
February 1, 2024 09:47
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
shahrs87
requested changes
Feb 2, 2024
| " already exists in state " + replicaInfo.getState() + | ||
| " and thus cannot be created."); | ||
| if (newGS != 0L) { | ||
| cleanupReplica(b.getBlockPoolId(), replicaInfo); |
There was a problem hiding this comment.
Add a comment on why are we cleaning up replica here.
|
@ritegarg There are still 45 test faiures in the latest run. Please take a look. |
|
@ayushtkn @Hexiaoqiao Can you guys please provide some early feedback? Thank you. |
Updated |
Fixed locally, started a new Jenkins build |
|
🎊 +1 overall
This message was automatically generated. |
shahrs87
requested changes
Feb 5, 2024
| */ | ||
| private void setupPipelineForAppendOrRecovery() throws IOException { | ||
| private boolean setupPipelineForAppendOrRecovery() throws IOException { |
There was a problem hiding this comment.
|
|
||
| // During create stage, if we remove a node (nodes.length - 1) | ||
| // min replication should still be satisfied. | ||
| if (isCreateStage && !(dfsClient.dtpReplaceDatanodeOnFailureReplication > 0 && |
There was a problem hiding this comment.
Reason behind adding this check here:
We are already doing this check in catch block of addDatanode2ExistingPipeline method here.
But when isAppend flag is set to false and we are in PIPELINE_SETUP_CREATE phase, we exit early from addDatanode2ExistingPipeline method here
Irrespective of ReplaceDatanodeOnFailure policy, we will NEVER add a new datanode to the pipeline during PIPELINE_SETUP_CREATE stage and if removing one bad datanode is going to violate dfs.client.block.write.replace-datanode-on-failure.min-replication property then we should exit early.
There was a problem hiding this comment.
Thinking about the case where we are in PIPELINE_SETUP_CREATE stage but isAppend is set to true, then it will not exit early from addDatanode2ExistingPipeline method here
Assuming the replication factor is 3 and dfs.client.block.write.replace-datanode-on-failure.min-replication is set to 3 and there is 1 bad node in the pipeline and there are valid nodes in the cluster, this patch will return false early.
I think we should move this check after handleDatanodeReplacement method. @ritegarg
|
💔 -1 overall
This message was automatically generated. |
added 3 commits
February 13, 2024 14:02
|
Closing this PR in favor of https://github.com/apache/hadoop/pull/6556/files |
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.
Description of PR
Adding rack failure tolerance when creating a new file. During create file operation(see logs in jira), in case of bad rack, the create operation failed and hence the operation was not rack failure tolerant. This was not same as what happens during append phase where min replication factor is applicable.
Adding this update to fix this behavior to make the create phase on the lines of append phase.
How was this patch tested?
Added Test Cases for following scenarios:-
For code changes:
LICENSE,LICENSE-binary,NOTICE-binaryfiles?