[Fix] Rewind seekable streams before retrying #821
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
pietern
approved these changes
Nov 14, 2024
ksafonov-db
suggested changes
Nov 14, 2024
ksafonov-db
reviewed
Nov 14, 2024
There was a problem hiding this comment.
I'd personally prefer if we were rewinding stream only when we need to. @pietern WDYT?
|
@ksafonov-db I doubt it matters much. Regardless of the outcome of the operation, the caller will have to seek to a known location anyway if they want to keep the handle open in the first place. Performance impact on failure is negligible because seek only updates the location offset in the file handle and doesn't trigger I/O directly (not counting readahead that the OS might do, but we can ignore that here). I'll leave it to you to figure out if there is a case for pushing this down into a "pre-run" callback from a structural pov. |
|
If integration tests don't run automatically, an authorized user can run them manually by following the instructions below: Trigger: Inputs:
Checks will be approved automatically on success. |
|
Test Details: go/deco-tests/11844787622 |
|
Confirmed with @ksafonov-db offline that he is good with the current state of the PR |
renaudhartert-db
added a commit
that referenced
this pull request
Nov 18, 2024
### New Features and Improvements * Read streams by 1MB chunks by default. ([#817](#817)). ### Bug Fixes * Rewind seekable streams before retrying ([#821](#821)). ### Internal Changes * Reformat SDK with YAPF 0.43. ([#822](#822)). * Update Jobs GetRun API to support paginated responses for jobs and ForEach tasks ([#819](#819)). * Update PR template ([#814](#814)). ### API Changes: * Added `databricks.sdk.service.apps`, `databricks.sdk.service.billing`, `databricks.sdk.service.catalog`, `databricks.sdk.service.compute`, `databricks.sdk.service.dashboards`, `databricks.sdk.service.files`, `databricks.sdk.service.iam`, `databricks.sdk.service.jobs`, `databricks.sdk.service.marketplace`, `databricks.sdk.service.ml`, `databricks.sdk.service.oauth2`, `databricks.sdk.service.pipelines`, `databricks.sdk.service.provisioning`, `databricks.sdk.service.serving`, `databricks.sdk.service.settings`, `databricks.sdk.service.sharing`, `databricks.sdk.service.sql`, `databricks.sdk.service.vectorsearch` and `databricks.sdk.service.workspace` packages. OpenAPI SHA: 2035bf5234753adfd080a79bff325dd4a5b90bc2, Date: 2024-11-15
This was referenced Nov 18, 2024
github-merge-queue bot
pushed a commit
that referenced
this pull request
Nov 18, 2024
### New Features and Improvements * Read streams by 1MB chunks by default. ([#817](#817)). ### Bug Fixes * Rewind seekable streams before retrying ([#821](#821)). * Properly serialize nested data classes. ### Internal Changes * Reformat SDK with YAPF 0.43. ([#822](#822)). * Update Jobs GetRun API to support paginated responses for jobs and ForEach tasks ([#819](#819)). ### API Changes: * Added `service_principal_client_id` field for `databricks.sdk.service.apps.App`. * Added `azure_service_principal`, `gcp_service_account_key` and `read_only` fields for `databricks.sdk.service.catalog.CreateCredentialRequest`. * Added `azure_service_principal`, `read_only` and `used_for_managed_storage` fields for `databricks.sdk.service.catalog.CredentialInfo`. * Added `omit_username` field for `databricks.sdk.service.catalog.ListTablesRequest`. * Added `azure_service_principal` and `read_only` fields for `databricks.sdk.service.catalog.UpdateCredentialRequest`. * Added `external_location_name`, `read_only` and `url` fields for `databricks.sdk.service.catalog.ValidateCredentialRequest`. * Added `is_dir` field for `databricks.sdk.service.catalog.ValidateCredentialResponse`. * Added `only` field for `databricks.sdk.service.jobs.RunNow`. * Added `restart_window` field for `databricks.sdk.service.pipelines.CreatePipeline`. * Added `restart_window` field for `databricks.sdk.service.pipelines.EditPipeline`. * Added `restart_window` field for `databricks.sdk.service.pipelines.PipelineSpec`. * Added `private_access_settings_id` field for `databricks.sdk.service.provisioning.UpdateWorkspaceRequest`. * Changed `create_credential()` and `generate_temporary_service_credential()` methods for [w.credentials](https://databricks-sdk-py.readthedocs.io/en/latest/workspace/credentials.html) workspace-level service with new required argument order. * Changed `access_connector_id` field for `databricks.sdk.service.catalog.AzureManagedIdentity` to be required. * Changed `access_connector_id` field for `databricks.sdk.service.catalog.AzureManagedIdentity` to be required. * Changed `name` field for `databricks.sdk.service.catalog.CreateCredentialRequest` to be required. * Changed `credential_name` field for `databricks.sdk.service.catalog.GenerateTemporaryServiceCredentialRequest` to be required. OpenAPI SHA: f2385add116e3716c8a90a0b68e204deb40f996c, Date: 2024-11-15
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.
What changes are proposed in this pull request?
This PR adapts the retry mechanism of
BaseClientto only retry if (i) the request is not a stream or (ii) the stream is seekable and can be reset to its initial position. This fixes a bug that led retries to ignore part of the request that were already processed in previous attempts.How is this tested?
Added unit tests to verify that (i) non-seekable streams are not retried, and (ii) seekable streams are properly reset before retrying.