Skip to content

DRIVERS-2888 Support QE with Client.bulkWrite #1770

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 19 commits into
base: master
Choose a base branch
from

Conversation

mdb-ad
Copy link
Contributor

@mdb-ad mdb-ad commented Mar 17, 2025

Add spec test for client bulkWrite QE.

C driver implementation

libmongocrypt requirement: 1.10
Drivers changes required:

Please complete the following before merging:

  • Update changelog.
  • Test changes in at least one language driver. (C driver)
  • Test these changes against all server versions and topologies (including standalone, replica set, sharded
    clusters, and serverless). (Note: C driver doesn't currently test CSFLE with sharded)

@mdb-ad mdb-ad marked this pull request as ready for review April 18, 2025 18:00
@mdb-ad mdb-ad requested a review from a team as a code owner April 18, 2025 18:00
@mdb-ad mdb-ad requested review from nbbeeken and removed request for a team April 18, 2025 18:00
@kevinAlbs kevinAlbs self-requested a review April 18, 2025 18:05
Copy link
Contributor

@nbbeeken nbbeeken left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I may be out of the loop on some of the context, are drivers changes needed to get these passing? Is there a specific libmongocrypt version needed?

There seems to be more unified test changes in the C driver than are here, are those just extra test sync(s) or are they new tests?

Node.js Test Failures
  1) CRUD unified
       client bulkWrite with queryable encryption
         client bulkWrite QE replaceOne:
     MongoBulkWriteError: Document failed validation
      at LegacyOrderedBulkOperation.handleWriteError (src/bulk/common.ts:1224:13)
      at executeCommands (src/bulk/common.ts:587:19)
      at processTicksAndRejections (node:internal/process/task_queues:105:5)
      at async BulkWriteShimOperation.execute (src/bulk/common.ts:882:12)
      at async tryOperation (src/operations/execute_operation.ts:283:14)
      at async executeOperation (src/operations/execute_operation.ts:115:12)
      at async LegacyOrderedBulkOperation.execute (src/bulk/common.ts:1211:12)
      at async BulkWriteOperation.execute (src/operations/bulk_write.ts:60:12)
      at async InsertManyOperation.execute (src/operations/insert.ts:147:19)
      at async tryOperation (src/operations/execute_operation.ts:283:14)
      at async executeOperation (src/operations/execute_operation.ts:115:12)
      at async Collection.insertMany (src/collection.ts:307:12)
      at async insertMany (test/tools/unified-spec-runner/operations.ts:411:10)
      at async executeOperationAndCheck (test/tools/unified-spec-runner/operations.ts:1039:14)
      at async runUnifiedTest (test/tools/unified-spec-runner/runner.ts:226:9)
      at async Context.<anonymous> (test/tools/unified-spec-runner/runner.ts:336:13)

  2) CRUD unified
       client bulkWrite with queryable encryption
         client bulkWrite QE with multiple replace fails:
     AssertionError: Operation clientBulkWrite succeeded but was not supposed to
      at executeOperationAndCheck (test/tools/unified-spec-runner/operations.ts:1057:12)
      at processTicksAndRejections (node:internal/process/task_queues:105:5)
      at async runUnifiedTest (test/tools/unified-spec-runner/runner.ts:226:9)
      at async Context.<anonymous> (test/tools/unified-spec-runner/runner.ts:336:13)

Copy link
Contributor

@kevinAlbs kevinAlbs left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggest updating this PR description to give an overview of needed changes. I expect implementors will refer to this PR (and the C driver PR).

C driver implementation

Suggest linking to the C driver PR (once created) rather than the Evergreen patch. The patch appears to include unrelated spec tests.

@mdb-ad mdb-ad requested a review from kevinAlbs July 17, 2025 06:07
@mdb-ad mdb-ad requested a review from a team as a code owner July 23, 2025 23:25
@mdb-ad mdb-ad requested review from katcharov and removed request for a team July 23, 2025 23:25
@mdb-ad mdb-ad requested a review from kevinAlbs July 23, 2025 23:29
@mdb-ad
Copy link
Contributor Author

mdb-ad commented Jul 24, 2025

I may be out of the loop on some of the context, are drivers changes needed to get these passing? Is there a specific libmongocrypt version needed?

libmongocrypt 1.10 is required.

There seems to be more unified test changes in the C driver than are here, are those just extra test sync(s) or are they new tests?

Description updated: the linked C driver changes in the description are bulkWrite-related only

@mdb-ad mdb-ad requested a review from isabelatkinson July 24, 2025 19:32
@kevinAlbs kevinAlbs removed the request for review from katcharov July 25, 2025 11:58
Copy link
Contributor

@isabelatkinson isabelatkinson left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Tests pass in Rust. Draft PR: mongodb/mongo-rust-driver#1445

There are some updates needed in the bulk write spec to account for the different command-building logic. Here's a commit with proposed changes: isabelatkinson@a060234


Expect this to fail since encryption results in a document exceeding the `maxBsonObjectSize` limit.

7. Use MongoClient.bulkWrite to insert the following into `coll2`:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These two tests should note that an 8.0+ server is required for bulkWrite support

@@ -565,10 +565,13 @@ First, perform the setup.
2. Using `client`, drop and create the collection `db.coll` configured with the included JSON schema
[limits/limits-schema.json](../limits/limits-schema.json).

3. Using `client`, drop the collection `keyvault.datakeys`. Insert the document
3. Using `client`, drop and create the collection `db.coll2` configured with the included encryptedFields
[limits/limits-encryptedFields.json](../limits/limits-encryptedFields.json).
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm missing this file, can you add it to the PR?

runOnRequirements:
- minServerVersion: "8.0"
serverless: forbid # Serverless does not support bulkWrite: CLOUDP-256344.
csfle: true
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@baileympearson
Copy link
Contributor

@kevinAlbs @mdb-ad Is there any chance https://jira.mongodb.org/browse/DRIVERS-2859 can be completed before DRIVERS-2888? It makes much more sense to make the change in once place than making the changes in every driver. This is especially true because drivers supporting non-document sequences for client bulk write will be temporary until https://jira.mongodb.org/browse/DRIVERS-2859 is addressed. So making drivers implement the changes before https://jira.mongodb.org/browse/DRIVERS-2859 will cost extra work in every driver, extra work updating the spec to allow (or enforce) document sequences with bulk write after https://jira.mongodb.org/browse/DRIVERS-2859 is implemented, and then again extra work in drivers to support document sequencing for bulkWrite with QE.

From a more selfish, Node-specific perspective - our bulkWrite implementation is tightly coupled to document sequences and making this change might be non-trivial. I'd rather not refactor our bulk write implementation unless absolutely necessary.

@kevinAlbs
Copy link
Contributor

drivers supporting non-document sequences for client bulk write will be temporary until https://jira.mongodb.org/browse/DRIVERS-2859 is addressed

The C driver only required a minor change to support QE with MongoClient.bulkWrite. Supporting CSFLE with MongoCollection.bulkWrite already requires passing all write documents in a non-document sequence, and the C driver reuses it.

I propose: proceed with this PR to unblock some drivers with QE, but add a blurb suggesting drivers wait for DRIVERS-2859:

Auto-Encryption

libmongocrypt does not-yet support OP_MSG document sequences (DRIVERS-2859). Drivers requiring significant changes to pass a bulkWrite command to libmongocrypt may prefer to wait until DRIVERS-2859 is implemented to support automatic encryption.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants