HDDS-15521. StreamBlockInputStream fails with TimeoutIOException without retry or failover#10479
Merged
Merged
Conversation
…out retry or datanode failover.
adoroszlai
reviewed
Jun 10, 2026
adoroszlai
left a comment
adoroszlai
left a comment
Contributor
There was a problem hiding this comment.
Thanks @sadanand48 for the patch. To reduce overhead of cluster startup, can you please rewrite test cases to share a single cluster?
adoroszlai
changed the title
amaliujia
reviewed
Jun 18, 2026
Contributor
There was a problem hiding this comment.
I didn't find enough context of this PR. Purely from I can tell so far, this PR try to address:
- The previous implementation do not catch and retry on TimeoutIOException
- The previous implementation do not retry for
streamingReader.read(length, preRead)which also could throw IO exception.
If I understand the goal correctly, I would suggest:
- Just simply add
if (cause instance of TimeoutIOException || cause instanceof StorageContainerException || isConnectivityIssue(cause))
- Do a reasonable while loop retry for
streamingReader.read(length, preRead) - The retry count for
initialize()andstreamingReader.read(length, preRead)should be shared.
szetszwo
reviewed
Jun 25, 2026
szetszwo
left a comment
szetszwo
left a comment
Contributor
There was a problem hiding this comment.
@sadanand48 , thanks for working on this! Just have some minor comments inlined.
szetszwo
approved these changes
Jun 29, 2026
szetszwo
left a comment
szetszwo
left a comment
Contributor
There was a problem hiding this comment.
+1 the change looks good.
Contributor
Author
|
Thanks @szetszwo , @amaliujia, @adoroszlai for the reviews |
errose28
added a commit
to errose28/ozone
that referenced
this pull request
Jul 7, 2026
* master: (519 commits) HDDS-14544. OM DB Insights: Duplicate API calls triggered when changing limit selector (apache#10677). HDDS-15587. [Recon] Show 0 for offline DN pending deletion instead of -1 (apache#10585). HDDS-15521. StreamBlockInputStream fails with TimeoutIOException without retry or failover. (apache#10479) HDDS-15170. Add mock-based unit tests for DataStream write path (apache#10230) HDDS-15552. Ratis events should not be published as metrics (apache#10523) HDDS-15746. Bump kerby to 2.1.2 (apache#10666) HDDS-15579. Replace SimpleSpanProcessor with BatchSpanProcessor (apache#10569) HDDS-15747. Address review comments for HDDS-15083 (apache#10669) HDDS-15732. Some ozone commands ignore OZONE_MODULE_ACCESS_ARGS (apache#10655) HDDS-15605. Fix flaky testContainerExclusionWithClosedContainerException (apache#10621) HDDS-11855. Fix flaky TestContainerBalancerDatanodeNodeLimit#checkIterationResultException (apache#10667) HDDS-15741. Bump awssdk to 2.46.17 (apache#10661) HDDS-10307. Speed up TestOzoneManagerHAWithStoppedNodes (apache#10658) HDDS-15651. Test case for DiskBalancer when markContainerForDelete fails (apache#10593) HDDS-15742. Bump nimbus-jose-jwt to 10.9.1 (apache#10662) HDDS-15719. Add check for allowed action usage in workflows (apache#10641) HDDS-15744. Bump javassist to 3.32.0-GA (apache#10665) HDDS-15737. Fix intermittent failure in balancerShouldOnlySelectConfiguredIncludeContainers (apache#10660) HDDS-15743. Bump gson to 2.14.0 (apache#10664) HDDS-11093. Fix intermittent failure in TestContainerBalancerDatanodeNodeLimit#testMetrics (apache#10659) ... Conflicts: hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/ozoneimpl/ContainerScanHelper.java
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
4 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.
What changes were proposed in this pull request?
StreamBlockInputStream: Added a retry loop in dataAvailableToRead() that catches read failures (including TimeoutIOException) and calls handleExceptions() to release the stream, refresh pipeline/token, and reset requestedLength = position so reads resume from the current offset. On retry, failed datanodes are recorded in a per-stream failedStreamingDatanodes set and excluded from the next init.
XceiverClientGrpc: Extended initStreamRead() to accept an excluded-datanode set and skip those nodes when opening a new streaming stream, so retries target a different replica instead of rebinding to the same dead one. Added INFO logging to show which datanode each init selects.
What is the link to the Apache JIRA
https://issues.apache.org/jira/browse/HDDS-15521
How was this patch tested?
Mini-Ozone cluster test added