docs: add resource warnings for embedding schema alter

docs-site/content/0.25.0/api/collections.md

@@ -1030,9 +1030,15 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
-If you need to do zero-downtime schema changes, you could also re-create the collection fully and use 
+If you need to do zero-downtime schema changes, you could also re-create the collection fully and use
 the [Collection Alias](./collection-alias.md) feature to do a zero-downtime switch over to the new collection:
 
 1. [Create your collection](#create-a-collection) as usual with a timestamped name. For eg: `movies_jan_1`

docs-site/content/0.25.0/api/vector-search.md

@@ -350,9 +350,25 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
-To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field. 
+To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.
 
 Here's an example:
 

docs-site/content/0.25.1/api/collections.md

@@ -1084,9 +1084,15 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
-If you need to do zero-downtime schema changes, you could also re-create the collection fully and use 
+If you need to do zero-downtime schema changes, you could also re-create the collection fully and use
 the [Collection Alias](./collection-alias.md) feature to do a zero-downtime switch over to the new collection:
 
 1. [Create your collection](#create-a-collection) as usual with a timestamped name. For eg: `movies_jan_1`

docs-site/content/0.25.1/api/vector-search.md

@@ -350,9 +350,25 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
-To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field. 
+To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.
 
 Here's an example:
 

docs-site/content/0.25.2/api/collections.md

@@ -1090,9 +1090,15 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
-If you need to do zero-downtime schema changes, you could also re-create the collection fully and use 
+If you need to do zero-downtime schema changes, you could also re-create the collection fully and use
 the [Collection Alias](./collection-alias.md) feature to do a zero-downtime switch over to the new collection:
 
 1. [Create your collection](#create-a-collection) as usual with a timestamped name. For eg: `movies_jan_1`

docs-site/content/0.25.2/api/vector-search.md

@@ -350,9 +350,25 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
-To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field. 
+To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.
 
 Here's an example:
 

docs-site/content/26.0/api/collections.md

@@ -1156,9 +1156,15 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
-If you need to do zero-downtime schema changes, you could also re-create the collection fully with the updated schema and use 
+If you need to do zero-downtime schema changes, you could also re-create the collection fully with the updated schema and use
 the [Collection Alias](./collection-alias.md) feature to do a zero-downtime switch over to the new collection:
 
 Let's say you have a collection called `movies_jan_1` that you want to change the schema for.

docs-site/content/26.0/api/vector-search.md

@@ -350,9 +350,25 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
-To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field. 
+To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.
 
 Here's an example:
 

docs-site/content/27.0/api/collections.md

@@ -1225,6 +1225,12 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
 If you need to do zero-downtime schema changes, you could also re-create the collection fully with the updated schema and use

docs-site/content/27.0/api/vector-search.md

@@ -350,9 +350,25 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
-To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field. 
+To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.
 
 Here's an example:
 

docs-site/content/27.1/api/collections.md

@@ -1225,6 +1225,12 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Using an alias
 
 If you need to do zero-downtime schema changes, you could also re-create the collection fully with the updated schema and use

docs-site/content/27.1/api/vector-search.md

@@ -395,6 +395,22 @@ To simplify the process of embedding generation, Typesense can automatically use
 
 When you do a search query on this automatically-generated vector field, your search query will be vectorized using the same model used for the field, which then allows you to do semantic search or combine keyword and semantic search to do hybrid search.
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
 To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.

docs-site/content/28.0/api/collections.md

@@ -1313,6 +1313,12 @@ curl "http://localhost:8108/collections/companies" \
        }'
 ```
 
+:::danger Adding Embedding Fields
+If you use this alter operation to add an [auto-embedding field](./vector-search.html#option-b-auto-embedding-generation-within-typesense), Typesense will generate embeddings for **all existing documents** in the collection. This is highly CPU and RAM intensive and **can make your cluster nodes unresponsive**.
+
+For production clusters, we strongly recommend creating a new collection with the embedding field in the schema and indexing documents in controlled batches, rather than altering an existing collection. See the [alias feature](#using-an-alias) for zero-downtime migration, and [GPU Acceleration](./vector-search.html#using-a-gpu-optional) to speed up embedding generation.
+:::
+
 ### Get Schema Change Status
 
 You can check the status of in-progress schema change operations by using the schema changes endpoint.

docs-site/content/28.0/api/vector-search.md

@@ -399,6 +399,22 @@ When you do a search query on this automatically-generated vector field, your se
 Embeddings are only regenerated when one or more fields specified in the `embed.from` configuration are updated. This helps avoid unnecessary embedding recreation and API calls when other fields in the document are modified.
 :::
 
+:::warning Adding Embeddings to Existing Collections
+When you add an auto-embedding field to an **existing collection** via [schema alter](./collections.html#update-or-alter-a-collection) (PATCH), Typesense will generate embeddings for **all existing documents at once**. This is computationally intensive and can cause:
+
+- **CPU spikes to 100%**, making nodes unresponsive to API calls
+- **High RAM usage**, as the cluster must have enough memory to hold all document embeddings
+- **Blocked writes** to the collection until the operation completes
+
+**Recommended approach for production clusters:**
+
+1. Create a **new collection** with the embedding field already in the schema
+2. Index documents into the new collection in a controlled manner (e.g., in batches)
+3. Use the [Collection Alias](./collections.html#using-an-alias) feature to switch traffic to the new collection with zero downtime
+
+If you must use schema alter, do so during off-peak hours on a cluster with sufficient CPU and RAM headroom. Enabling [GPU Acceleration](#using-a-gpu-optional) significantly speeds up embedding generation.
+:::
+
 ### Creating an auto-embedding field
 
 To create a field that automatically embeds other string or string array fields, you need to set the `embed` property of the field.