The cosmosSearch.query is not enabled for this cluster

Question

The cosmosSearch.query is not enabled for this cluster

Julius Dietmar 60

I have a M30 DocumentDB cluster (which, according to the documentation here: https://devblogs.microsoft.com/cosmosdb/diskann-in-azure-cosmos-db-for-mongodb supports DiskANN vector search) and created successfully vector search indexes on my items collection. However, when I execute a search query, I receive this error:

Index creation:

  "createIndexes": "items",
  "indexes": [
    {
      "name": "text_vector_index",
      "key": { "textEmbeddings": "cosmosSearch" }, // Path to your vector array field
      "cosmosSearchOptions": {
        "kind": "vector-diskann",               // Options: 'vector-hnsw', 'vector-ivf', or 'vector-diskann'
        "dimensions": 1536,                  // Vector dimensions (e.g., 1536 for OpenAI embeddings)
        "similarity": "COS",                 // Options: 'COS' (Cosine), 'L2' (Euclidean), 'IP' (Dot Product)
        "maxDegree": 64,
        "lBuild": 128
      }
    }
  ]
})

Query:

db.items.aggregate([ { $search: { cosmosSearch: { path: "textEmbeddings", query: [0.1, ... 0.2], k: 1 } } } ])

MongoServerError[CommandNotSupported]:

Do I have to enable it and if so, where? If not, why is it not enabled?

SAI JAGADEESH KUDIPUDI 3,645 Reputation points Microsoft External Staff Moderator

2026-07-22T02:13:02.8466667+00:00

Hi @Julius Dietmar ,

Thank you for reaching out and for sharing the index definition and error details.

The error:

"The cosmosSearch.query is not enabled for this cluster"

occurs because your cluster is running on the M30 tier, while your vector index is configured with:"kind": "vector-diskann"

According to the Azure Cosmos DB for MongoDB (vCore) limits documentation, DiskANN and HNSW vector indexes are supported only on M40 and higher cluster tiers.

Although the createIndexes command may succeed on an M30 cluster, the query engine cannot execute DiskANN-based vector searches on that tier. As a result, when the $search stage attempts to use the DiskANN index, the service returns:

The cosmosSearch.query is not enabled for this cluster Upgrade to M40 or higher (Recommended)

If you require DiskANN or HNSW vector search capabilities, scale the cluster to M40 or above and rerun the query. Recreating the index may also be required after the upgrade.

Reference: Azure Cosmos DB for MongoDB (vCore) Vector Search Limits

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

Answer accepted by question author

0 additional answers

Your answer

SAI JAGADEESH KUDIPUDI 3,645 Reputation points Microsoft External Staff Moderator

2026-07-22T02:13:02.8466667+00:00

Hi @Julius Dietmar ,

Thank you for reaching out and for sharing the index definition and error details.

The error:

"The cosmosSearch.query is not enabled for this cluster"

occurs because your cluster is running on the M30 tier, while your vector index is configured with:"kind": "vector-diskann"

According to the Azure Cosmos DB for MongoDB (vCore) limits documentation, DiskANN and HNSW vector indexes are supported only on M40 and higher cluster tiers.

Although the createIndexes command may succeed on an M30 cluster, the query engine cannot execute DiskANN-based vector searches on that tier. As a result, when the $search stage attempts to use the DiskANN index, the service returns:

The cosmosSearch.query is not enabled for this cluster Upgrade to M40 or higher (Recommended)

If you require DiskANN or HNSW vector search capabilities, scale the cluster to M40 or above and rerun the query. Recreating the index may also be required after the upgrade.

Reference: Azure Cosmos DB for MongoDB (vCore) Vector Search Limits

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

Answer 1

Hello Julius,

M30 supports DiskANN, so you shouldn’t need to enable cosmosSearch manually. The index definition looks valid, but the query uses query; the current syntax expects the embedding under vector.

Try:

db.items.aggregate([
  {
     $search: {
      cosmosSearch: {
        path: "textEmbeddings",
        vector: [0.1, 0.2 /* full 1536-value vector */],
        k: 1,
        lSearch: 40
       }
     }
   }
])

Also confirm that the query contains exactly 1,536 numeric values, matching the index dimensions, and check the index status with db.items.getIndexes().

If the corrected query still returns “cosmosSearch query is not enabled for this cluster,” the feature may not have been provisioned correctly on that specific cluster. At that point, recreate the M30 cluster or open an Azure support request with the cluster resource ID and region, because this would be a service-side enablement issue rather than something exposed as a portal setting. Microsoft’s current documentation confirms DiskANN and cosmosSearch support on M30 and higher tiers.

Please "Accept the answer" if this information helped you. This will help us and others in the community as well.