Steps to reproduce

Title

Retry schedule exits immediately on FAILED_TO_START_DUE_TO_NO_CAPACITY

Description

When a run with a retry schedule fails to start due to no available capacity, it exits immediately with RETRY_LIMIT_EXCEEDED instead of honoring the retry duration.

Steps to Reproduce

1. Create a run with a retry schedule (e.g., 2 hours) and non existing hardware specs.
2. Schedule the run for a future time (in UTC)
3. Job fails with FAILED_TO_START_DUE_TO_NO_CAPACITY
4. Run immediately terminates with RETRY_LIMIT_EXCEEDED

Expected Behavior

The run should retry for the configured duration (2 hours) when capacity becomes available.

Actual Behavior

Run exits immediately without retrying.

Logs

02/25/2026 07:20 Run main/kimi-k25 Run status changed PENDING -> SUBMITTED
02/25/2026 07:20 Job main/kimi-k25-0-0 Job status changed SUBMITTED -> TERMINATING. Termination reason: FAILED_TO_START_DUE_TO_NO_CAPACITY
02/25/2026 07:20 Job main/kimi-k25-0-0 Job status changed TERMINATING -> FAILED
02/25/2026 07:20 Run main/kimi-k25 Run status changed SUBMITTED -> TERMINATING. Termination reason: RETRY_LIMIT_EXCEEDED
02/25/2026 07:20 Run main/kimi-k25 Run status changed TERMINATING -> PENDING

Environment

• dstack version: 0.20.10
• Deployment: [self-hosted/kubernetes]

Actual behaviour

No response

Expected behaviour

No response

dstack version

0.20.10

Server logs

Additional information

No response