Cosmos Partition merge requirements

Jesse 20 Reputation points
2026-09-28T04:35:52.99+00:00

I'm trying to determine if our cosmos db is eligible for partition merge, however when i follow the instructions here: https://learn.microsoft.com/en-us/azure/cosmos-db/merge?tabs=azure-powershell%2Cnosql

It takes me to a page with some AI generated content and not the table with the "Pass" next to the criteria. Has this been moved? How do i know if the db is eligible?

I also had another question regarding how the partition merge works. When enabled does it automatically do the merge or do i need to start it somewhere? Also does this disable the db so no read/writes can happen while it merges?

Azure Cosmos DB
Azure Cosmos DB

An Azure NoSQL database service for app development.


1 answer

Sort by: Oldest
  1. SHOUMIK CHAKRAVARTY 910 Reputation points
    2026-09-29T03:58:44.8666667+00:00

    Hello @Jesse - The criteria are still on that page, they just moved. They are now under Limitations in a section called “Preview eligibility criteria,” rather than near the top where the old table was. https://learn.microsoft.com/en-us/azure/cosmos-db/merge?tabs=azure-powershell%2Cnosql#preview-eligibility-criteria

    For your actual question, there is a better route than reading the list. The page documents a portal checker:

    "To check whether an Azure Cosmos DB account is eligible for the preview, you can use the built-in eligibility checker in the Azure portal. From your Azure Cosmos DB account overview page in the Azure portal, navigate to Diagnose and solve problems -> Throughput and Scaling -> Partition Merge. Run the Check eligibility for partition merge preview diagnostic."

    That gives you a definitive answer for your account. For reference, the criteria are API for NoSQL or MongoDB 3.6+, provisioned throughput rather than serverless, single write region, and not using point-in-time restore, customer-managed keys, per-partition automatic failover, vector indexing and search, or analytical store for Synapse Link.

    On whether it merges automatically: no, enabling the feature does not start anything. You enable it on the Features page, wait 15 to 20 minutes for it to take effect, then run the merge yourself through PowerShell or the CLI. PowerShell supports a -WhatIf flag that simulates the merge and returns the expected result without executing it, which is worth running first.

    To know when it is finished, the page suggests checking the Activity Log and filtering for “Merge the physical partitions of a SQL container” or the MongoDB equivalent.

    On read and write availability, the merge itself is not documented as taking the database offline. But there is a much bigger availability catch in enabling the feature at all, and it is worth reading before you turn it on:

    "When merge is enabled on an account, only requests from .NET SDK version >= 3.27.0 or Java SDK >= 4.42.0 or Azure Cosmos DB Spark connector >= 4.18.0 or JavaScript SDK >= 4.3 will be allowed on the account, regardless of whether merges are ongoing or not. Requests from other SDKs (older .NET SDK, older Java SDK, older JavaScript SDK, any Python SDK, any Go SDK) or unsupported connectors (Azure Data Factory, Azure Search, and Azure Functions extension <= 3.x) will be blocked and fail."

    So any Python or Go client is blocked outright for as long as the feature is enabled, not just during a merge. This applies to API for NoSQL. The page notes that there are no SDK or driver requirements for API for MongoDB.

    Turning it back off is not instant either. The page says it may take 15 to 20 minutes before requests from SDKs and connectors that are not supported for merge are allowed again.

    Help make this community better for everyone: if this answer resolved your issue, please accept it or leave an upvote. If not, share more details in a comment so we can continue the discussion and find the right solution.

    Was this answer helpful?


Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.