Skip to content

update(daytona): Reduces autoStopInterval globally and in analyzer - #1188

Open
dalinkstone wants to merge 1 commit into
abundant-ai:stagingfrom
dalinkstone:fix/daytona-auto-stop
Open

update(daytona): Reduces autoStopInterval globally and in analyzer#1188
dalinkstone wants to merge 1 commit into
abundant-ai:stagingfrom
dalinkstone:fix/daytona-auto-stop

Conversation

@dalinkstone

@dalinkstone dalinkstone commented Aug 11, 2026

Copy link
Copy Markdown

Reduce autoStopInterval globally and in analyzer

Description

The core issue happening due to a longer autoStopInterval is that sandbox concurrency can consume quotas at faster rates than normal. Generally an autoStopInterval of 30 minutes is used broadly, but for the usecase with oddish it is too small and can cause issues. An interval of 60 minutes may be possible, but to eliminate some simple quota consumption we use the 90 minute interval to avoid immediate errors.


Note

Medium Risk
Changes sandbox lifecycle timing for Harbor and analyzer paths, which can affect orphan cleanup, quota usage, and premature teardown if idle windows are too short.

Overview
Shortens Daytona sandbox idle auto-stop to reduce quota burn from orphaned sandboxes, and gives analyzer sandboxes their own shorter backstop.

Harbor trials drop daytona_auto_stop_interval_mins from 120 → 90.

Analyzer sandboxes no longer inherit that long Harbor interval. analyzer_block_runner stops passing the global value, and provisioning falls back to a new analyzer_daytona_auto_stop_interval_mins setting (15 by default). Explicit SandboxConfig.auto_stop_minutes overrides are still respected, including distinguishing None from a set value.

Reviewed by Cursor Bugbot for commit 3aa2d4e. Bugbot is set up for automated code reviews on this repo. Configure here.

…15 minutes in the analyzer

The core issue happening due to a longer autoStopInterval is that sandbox concurrency can consume quotas at faster rates than normal. Generally an autoStopInterval of 30 minutes is used broadly, but for the usecase with oddish it is too small and can cause issues. An interval of 60 minutes may be possible, but to eliminate some simple quota consumption we use the 90 minute interval to avoid immediate errors.

Signed-off-by: Dalin Stone <dalinkstone@gmail.com>
@vercel

vercel Bot commented Aug 11, 2026

Copy link
Copy Markdown
Contributor

@dalinkstone is attempting to deploy a commit to the Abundant Team on Vercel.

A member of the Team first needs to authorize it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant