Improve Download Memory Usage #61
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
aajtodd
reviewed
Oct 7, 2024
| @@ -129,18 +129,30 @@ fn handle_discovery_chunk( | |||
|
|
|||
| /// Download operation specific state | |||
| #[derive(Debug)] | |||
| pub(crate) struct DownloadState {} | |||
| pub(crate) struct DownloadState { | |||
| current_seq: u64, | |||
There was a problem hiding this comment.
Can you not get away with an atomic here?
There was a problem hiding this comment.
Thanks, updated to AtomicU64.
| if let Some(stream) = initial_chunk { | ||
| let seq = handle.ctx.next_seq(); |
There was a problem hiding this comment.
start_seq and seq aren't connected anywhere here, should probably set start_seq to seq instead of hard coded to 1
There was a problem hiding this comment.
Thanks, we would have to set start_seq = seq+1. I have added a current_seq() function which will return 0 or 1 depending upon we called next_seq or not.
It didn’t help much since we were only doing 125 out of 3840 parts out of order. |
aajtodd
approved these changes
Oct 7, 2024
ysaito1001
approved these changes
Oct 7, 2024
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Outdated suggestions cannot be applied.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 changes:
Tokio.spawndoesn't respect the spawn order, which can result in us downloading the firstnum_concurrencyparts in random order. For a workload of 5GB * 100 files, this can lead to very high memory usage, as seen in the diagram below. This PR refactors the exact part to be determined only once the task has been scheduled.Uploads can also have a similar issue where we read too many parts into memory. To fix that, we will need to refactor our scheduler to be smarter so that we only read the part when we have the permit. (Created: #60)
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.