Skip to content

docs: update ExitMode docs for v2 rework #175

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jmhsieh merged 1 commit into from
Mar 3, 2026
Merged

docs: update ExitMode docs for v2 rework #175

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 5 additions & 16 deletions docs/geneva/jobs/contexts.mdx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,12 +99,12 @@ See the API docs for all the parameters [`GenevaCluster.create_kuberay()`](https

#### Exit Modes

By default, the KubeRay cluster waits for all running jobs to complete before deleting. You can customize this behavior with the `on_exit` parameter:
When you launch multiple async jobs in a single context, the exit mode controls whether the cluster waits for all of them to finish and how it handles failures. You can customize this behavior with the `on_exit` parameter:

```python Python icon="python"
from geneva.runners.ray.raycluster import ExitMode

with db.context(cluster=cluster_name, manifest=manifest_name, on_exit=ExitMode.DELETE_AFTER_JOBS):
with db.context(cluster=cluster_name, manifest=manifest_name, on_exit=ExitMode.DELETE):
fut1 = tbl.backfill_async("embedding_a")
fut2 = tbl.backfill_async("embedding_b")
# No need to call .result() — the context waits for both jobs
Expand All @@ -113,20 +113,9 @@ with db.context(cluster=cluster_name, manifest=manifest_name, on_exit=ExitMode.D

| Exit Mode | Behavior |
|-----------|----------|
| `ExitMode.DELETE_AFTER_JOBS` (default) | Wait for all async jobs to complete, then delete. Ideal for batch scripts using `backfill_async()`. |
| `ExitMode.DELETE` | Always delete the cluster immediately on exit, without waiting for running jobs. |
| `ExitMode.DELETE_ON_SUCCESS` | Delete on success; retain if an exception occurred. Useful for debugging. |
| `ExitMode.RETAIN` | Never delete the cluster. Useful for notebooks and interactive sessions. |

<Tip>
`DELETE_AFTER_JOBS` is the default — the cluster stays alive until all running jobs have finished, then cleans itself up automatically. You can set a `wait_timeout` (in seconds) to cap how long the cluster waits before deleting:

```python Python icon="python"
with db.context(cluster=cluster_name, on_exit=ExitMode.DELETE_AFTER_JOBS, wait_timeout=300):
fut = tbl.backfill_async("embedding")
# Waits up to 5 minutes for jobs, then deletes regardless
```
</Tip>
| `ExitMode.DELETE` (default) | Wait for all async jobs in the context to complete, then delete the cluster. Ideal for batch scripts that launch multiple `backfill_async()` calls in one context. |
| `ExitMode.RETAIN_ON_FAILURE` | Wait for all async jobs in the context to complete. If any job failed, the context body raised an exception, or `wait_timeout` was exceeded, retain the cluster for debugging; otherwise delete. |
| `ExitMode.RETAIN` | Never delete the cluster, regardless of job outcomes. Useful for notebooks and interactive sessions where you run multiple jobs over time. |

### External Ray cluster
If you already have a Ray cluster, Geneva can execute jobs against it too. You do so by defining a Geneva cluster with [`GenevaCluster.create_external()`](https://lancedb.github.io/geneva/api/cluster/#geneva.cluster.mgr.GenevaCluster.create_external) which has the address of the cluster. Here's an example:
Expand Down
Loading