Errors

Retry a flaky step

Rescue an error, retry a limited number of times, and always record how it ended.

Reference

Edit the source, then Run. Checks start after your first Run. Ctrl/⌘ + Enter runs; Tab indents. Preview services use sample data and never contact a real service.

Sourceupstream/errors/retry_on_failure.vibe
# Stands in for a call that fails until its third attempt.
def flaky_fetch(attempt: int, *, succeeds_on: int) -> string
  raise RuntimeError, "timeout on attempt #{attempt}" if attempt < succeeds_on
  "payload from attempt #{attempt}"
end

type Outcome = { result: string, attempts: int, log: array<string> }

def fetch_with_retries(*, succeeds_on: int, limit: int) -> Outcome
  attempts = 0
  log: array<string> = []
  result = begin
    attempts += 1
    flaky_fetch(attempts, succeeds_on: succeeds_on)
  rescue RuntimeError => error
    log << error.message
    retry if attempts < limit
    "gave up"
  ensure
    log << "finished after #{attempts} attempts"
  end

  { result: result, attempts: attempts, log: log }
end

def run -> { recovers: Outcome, gives_up: Outcome }
  {
    recovers: fetch_with_retries(succeeds_on: 3, limit: 5),
    gives_up: fetch_with_retries(succeeds_on: 10, limit: 2)
  }
end

Output

Click Run example to see the result.

Code checks

Run once to load the checker. Your code stays in this browser.