gcp,s3,azure: make the storage client upload chunk size configurable #80668
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 change is |
|
I wonder if it makes sense to be per-provider, or if we should just have a "cloudstorage.write_chunk_size" or something that everyone reads? I'm not really sure either way |
|
Good point, S3 and Azure both expose configurable chunk sizes. S3 defaults to 5MB, Azure to 4MB let's tie them all to this shared setting. |
adityamaru
force-pushed
the
chunk-size-drop
branch
from
6fda95f to
33d36a6
Compare
adityamaru
changed the title
|
Changed to a single, shared cluster setting. I've not marked the setting as public since we don't expect users to change this. |
This change adds a `cloudstorage.write_chunk_size` cluster setting that allows us to control the size of the chunks buffered by the cloud storage client when uploading a file to storage. The setting defaults to 8MiB. Prior to this change gcs used a 16MB buffer, s3 a 5MB buffer, and azure a 4MB buffer. A follow up change will add memory monitoring to each external storage writer to account for these buffered chunks during upload. This change was motivated by the fact that in google-cloud-storage SDK versions prior to v1.21.0 every chunk is given a hardcoded timeout of 32s to successfully upload to storage. This includes retries due to transient errors. If any chunk during a backup were to hit this timeout the entire backup would fail. We have additional work to do to make the job more resilient to such failures, but dropping the default chunk size might mean we see fewer chunks hit their timeouts. Release note: None
adityamaru
force-pushed
the
chunk-size-drop
branch
from
33d36a6 to
c5abb54
Compare
|
friendly ping! |
dt
approved these changes
May 3, 2022
|
TFTR! bors r=dt |
|
Build succeeded: |
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.
This change adds a
cloudstorage.write_chunk_sizecluster settingthat allows us to control the size of the chunks buffered by the
cloud storage client when uploading a file to storage. The setting defaults to
8MiB.
Prior to this change gcs used a 16MB buffer, s3 a 5MB buffer, and azure a 4MB
buffer. A follow up change will add memory monitoring to each external storage
writer to account for these buffered chunks during upload.
This change was motivated by the fact that in google-cloud-storage
SDK versions prior to v1.21.0 every chunk is given a hardcoded
timeout of 32s to successfully upload to storage. This includes retries
due to transient errors. If any chunk during a backup were to hit this
timeout the entire backup would fail. We have additional work to do
to make the job more resilient to such failures, but dropping the default
chunk size might mean we see fewer chunks hit their timeouts.
Release note: None