Add a node failure policy: retries, and what happens when retrying doesn't help - #354

Closed

Todd J. Green (tjgreen42) wants to merge 13 commits into

mainfrom

Commits on Aug 24, 2026

Add RetryPolicySpec to orchestration input
Todd J. Green (tjgreen42)
committed
f268821
Add max_attempts, max_backoff, and on_failure to df.start()
Todd J. Green (tjgreen42)
committed
674fa31
Thread the retry policy through subtree and loop generations
Todd J. Green (tjgreen42)
committed
3068711
Retry node activities and unwind exhausted nodes to the enclosing loop
Todd J. Green (tjgreen42)
committed
43073c5
Remove the 100,000-iteration loop cap
Todd J. Green (tjgreen42)
committed
57dd97b
Bump to 0.2.7 and add the 0.2.6 -> 0.2.7 upgrade script
Todd J. Green (tjgreen42)
committed
4aa6b97
Add an end-to-end test for the node failure policy
Todd J. Green (tjgreen42)
committed
65ff216
Document the node failure policy
Todd J. Green (tjgreen42)
committed
ee7ab2e
Fold implementation experience back into the failure policy spec
Todd J. Green (tjgreen42)
committed
d1e6593
Move the failure policy spec and plan to the docs/ naming convention
Todd J. Green (tjgreen42)
committed
7483f15
Make the node failure policy opt-in and skip unretryable errors
Todd J. Green (tjgreen42)
committed
d22bb74
Document the reworked failure policy and drop the implementation plan
Todd J. Green (tjgreen42)
committed
ede2351

Commits on Aug 25, 2026

Set out the alternatives for where the failure policy attaches
Todd J. Green (tjgreen42)
committed
37cda16