revert nio sam file merger #2333
Closed
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
This reverts commit e98fe09.
it seems like the issues related to some sort of race condition with caching and deleting/renaming files in the gcs hdfs adaptor changing the behavior so that temporary files are first written to a temporary directory and then concatenated into the final file previously it wrote them to a directory with the same name as the final file, renamed that, and then copied them into the renamed file improving exception message when failing to write a bam on spark
|
I think we should wait for the NIO-based fix rather than merging this, which would be useless for our performance work. |
|
@droazen I added a comment on #2287 about what's needed from the GCS NIO side (see also googleapis/google-cloud-java#1519). |
|
closing this then in favor of waiting for nio updates |
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Outdated suggestions cannot be applied.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.
@tomwhite @droazen This fixes the problem I was seeing, but it's insanely slow for some reason. A 30gb file on a heavy duty cluster was taking between 30-50 minutes depending on how I sharded it. Most of that time is spent on the merging step where the master concatenates the chunks.
I think the refactoring of the writing that I've done isn't a bad thing to have. I added some extra log statements that it much easier to understand what's going on when it's slow.