@kapa.ai Yesterday I had updated the resources for the mssql source connector using UPDATE actor_definition SET resource_requirements = '{ "jobSpecific": [ { "jobType": "sync", "resourceRequirements": { "cpu_limit": "0.5", "cpu_request": "0.5", "memory_limit": "10Gi", "memory_request": "1Gi" } } ]}' WHERE id = 'connector id'; and had confirmed that the resources had updated by running SELECT name, resource_requirements FROM actor_definition WHERE name LIKE '%MSSQL%'; Today when I run the SLEECT query again, I see the "resource_requirements" column empty. What is happening? I am on Airbyte 2.0

You’re not doing anything wrong with the SQL itself; the behavior you’re seeing is most likely due to how Airbyte 2.0 manages connector definitions and their

resource_requirements

internally, but the knowledge sources do not explicitly document this reset behavior, so I can’t state the exact mechanism with certainty. What is documented: • In 2.0,

actor_definition.resource_requirements

is still the place to set connector‑definition–level resources (for all connectors of that type). The recommended pattern is exactly what you used: an

UPDATE actor_definition SET resource_requirements = '{"jobSpecific": [...]}' ...

. [Configuring resources 2.0; actor_definition table] • There is also a higher‑precedence, connection‑specific override via the

connection.resource_requirements

column. [Connection-specific config] However, none of the provided docs or issues explain why

actor_definition.resource_requirements

would appear set one day and then be empty the next. The sources do not mention any automatic cleanup, migration, or overwrite process in 2.0 that would clear that column. So I can’t definitively say what cleared it. Given the limitations of the docs, here’s what you can reliably do now, based only on the sources: 1. Re‑apply the connector‑definition resources Run your

UPDATE actor_definition ...

again and confirm with:

SELECT name, resource_requirements
   FROM actor_definition
   WHERE name LIKE '%MSSQL%';

This is the documented way to set definition‑level resources. [Configuring resources 2.0] 2. Optionally set connection‑specific resources (highest precedence) If you want to be sure the MSSQL syncs keep using your desired limits even if definition‑level settings change, set them at the connection level:

UPDATE connection
   SET resource_requirements = '{"cpu_limit": "0.5", "cpu_request": "0.5", "memory_limit": "10Gi", "memory_request": "1Gi"}'
   WHERE id = '<connection-id>';

[Connection-specific config] 3. Verify what the jobs actually use For a new sync, check the job logs and confirm the merged resource requirements being applied to the MSSQL source container. Airbyte logs the effective resources for each container when launching the replication pod. [Debugging resources] Because the knowledge sources don’t describe any 2.0 behavior that would clear

actor_definition.resource_requirements

overnight, I can’t explain the exact cause from documentation alone. If this keeps happening after you re‑apply the settings, it’s likely an implementation bug or some process overwriting definitions, and you’d need to check with Airbyte support or GitHub issues with your exact version and deployment details. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai the mssql connector v4.3.4 in airbyte 2.0 does not seem to have options to set batch size, concurrency, etc. What is the stable version of the mssql connector with these parameters?

The knowledge sources don’t list MSSQL source connector versions together with their configuration options (batch size, concurrency, etc.), so I can’t reliably say which exact version first/stably exposes those parameters in the UI or spec. What I can see: • The MSSQL source configuration (as exposed via the Terraform provider) includes: ◦

concurrency

– “Maximum number of concurrent queries to the database.” ◦

checkpoint_target_interval_seconds

◦

jdbc_url_params

and other fields. [Source MSSQL config] This confirms that a recent MSSQL source does support a

concurrency

parameter, but the docs don’t tie this schema to a specific connector version number. • The MSSQL destination changelog is detailed, but that’s a different connector (destination, not source) and its versions (e.g., 2.x.y, 1.x.y, 0.1.x) don’t apply to the source you’re asking about. [MSSQL destination changelog] • The MSSQL source migration guide only documents behavioral changes for versions 2.0.0, 3.0.0, and 4.0.0 (CDC cursor, datetime type remapping, large-table incremental), and does not mention batch size or concurrency options being added/removed in any specific version. [MSSQL source migrations] Because of this, I cannot confidently identify a “stable version” of the MSSQL source connector that you should downgrade/upgrade to in order to get batch size and concurrency options in the UI. What you can do with the information we do have: • If you’re on Airbyte 2.0 and using MSSQL source v4.3.4, but don’t see concurrency in the UI, you can still configure it via the API/Terraform using the

concurrency

field in the source configuration JSON, since that field is part of the documented schema. [Source MSSQL config] • For batch size specifically, the provided docs do not mention any batch-size parameter for the MSSQL source at all, so I can’t confirm that any version exposes it. Given the gaps in the docs, the safest next step is: 1. Inspect the current source spec via the Airbyte API (

/source_definition_specifications/get

) for your MSSQL source definition to see which fields are actually supported in v4.3.4. 2. If those fields are missing and you need a version that has them, check the connector’s GitHub changelog or open an issue, since the provided knowledge sources don’t map versions to those options. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.