Use withRetry in GpuCoalesceBatches #7852
Merged
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
Signed-off-by: Alessandro Bellina <[email protected]>
|
build |
1 similar comment
|
build |
|
build |
revans2
reviewed
Mar 7, 2023
sql-plugin/src/main/scala/com/nvidia/spark/rapids/RmmRapidsRetryIterator.scala
Outdated
Show resolved
Hide resolved
sql-plugin/src/main/scala/com/nvidia/spark/rapids/GpuCoalesceBatches.scala
Outdated
Show resolved
Hide resolved
revans2
previously approved these changes
Mar 7, 2023
sql-plugin/src/main/scala/com/nvidia/spark/rapids/RmmRapidsRetryIterator.scala
Show resolved
Hide resolved
sql-plugin/src/main/scala/com/nvidia/spark/rapids/GpuCoalesceBatches.scala
Show resolved
Hide resolved
|
build |
revans2
previously approved these changes
Mar 7, 2023
|
I am looking into the build failure, it is the smoke test for UCX. |
|
So what is happening in the smoke test is we fail to come up with UCX because we fail to allocate bounce buffers. The cause is likely that a previous test is still on the GPU, taking all the memory and causing this failure. I don't have a solution yet... |
abellina
requested review from
jlowe,
tgravescs,
GaryShen2008,
NvTimLiu and
zhanga5
as code owners
|
build |
revans2
reviewed
Mar 8, 2023
|
build |
|
build |
|
@revans2 this should be ready to go in. Run is clean, and all the directories show no leaks or double closes. nvidia-smi shows that we ended up in a V100 @ 32GB so I think the multiple app theory causing a UCX OOM holds. |
revans2
approved these changes
Mar 8, 2023
sameerz
added
the
reliability
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Outdated suggestions cannot be applied.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.
Closes #7777
Closes #7855
This adds retry/retry+split semantics to the GpuCoalesceBatches iterators (two of them). It does not handle
HostToGpuCoalesceIteratorbecause the column builder code in cuDF needs some changes for it to be retriable. Note #7851.This introduces a
withRetryNoSplitthat doesn't take any arguments, it's simply a retry block:Additionally, this PR adjusts the parallelism for the UCX shuffle smoke test to 1, so we don't run the risk of OOMing.