# Compute Pool Setup

> Guide the user through creating and configuring a Snowpark Container Services compute pool for the migration orchestrator.

- Skill: `snowflake-labs/compute-pool-setup` (Agent Skill)
- Install (CLI): `npx skillmds@latest add snowflake-labs/compute-pool-setup`
- Raw SKILL.md: https://api.skillmd.com/api/skills/snowflake-labs/compute-pool-setup/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- License: Proprietary. See License-Skills for complete terms
- Author: snowflake-labs (https://skillmd.com/u/snowflake-labs)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/snowflake-labs/compute-pool-setup

---


# Compute Pool Setup

Create a compute pool for the migration orchestrator to run inside Snowpark Container Services.

## Step 1 — Choose a Name

Ask the user:

> What name would you like for the compute pool? (e.g., `MIGRATION_COMPUTE_POOL`)

## Step 2 — Create the Compute Pool

Run the following SQL, substituting the user's chosen name:

```sql
CREATE COMPUTE POOL <COMPUTE_POOL_NAME>
  MIN_NODES = 1
  MAX_NODES = 1
  INSTANCE_FAMILY = CPU_X64_S
  AUTO_SUSPEND_SECS = 60;
```

> **Defaults:** `CPU_X64_S` is sufficient for the orchestrator. The orchestrator itself is lightweight — it coordinates work but does not process data. `MIN_NODES = 1` and `MAX_NODES = 1` are recommended; the pool scales the number of nodes, not the workers.
>
> **`AUTO_SUSPEND_SECS = 60`** lets the pool go cold ~60s after the orchestrator service suspends, so an idle pool stops billing even if teardown is never run. It stays active while the service is running, and `data_infrastructure(mode="up")` resumes it before the next wave (a ~30–60s warm-up). Raise it if your migration waves run back-to-back and the warm-up is disruptive.

## Step 3 — Verify the Pool Is Active

Run:

```sql
DESCRIBE COMPUTE POOL <COMPUTE_POOL_NAME>;
```

Check that the `state` column is **ACTIVE** or **IDLE**. If the state is `STARTING`, wait a moment and re-check — pools typically become active within a minute.

If the state is `SUSPENDED`, run:

```sql
ALTER COMPUTE POOL <COMPUTE_POOL_NAME> RESUME;
```

Then re-verify.

## Step 4 — Grant USAGE to the Migration Role

The role that will run the migration needs USAGE on the compute pool:

```sql
GRANT USAGE ON COMPUTE POOL <COMPUTE_POOL_NAME> TO ROLE <MIGRATION_ROLE>;
```

## Done

Return to `../SKILL.md`; the compute pool name is passed to `data_infrastructure(mode="up", compute_pool="<COMPUTE_POOL_NAME>")` when bringing the shared infrastructure up (it is persisted for reuse).