@cfxlabsinc/b2b-services
    Preparing search index...

    Variable DEFAULT_VICTOR_BACKFILL_SLICE_TIMEOUTConst

    DEFAULT_VICTOR_BACKFILL_SLICE_TIMEOUT: "1 hour"

    Ceiling on one backfill child run (workflowExecutionTimeout). The sync workflow's activity retry policy is unbounded — correct on the 30-minute schedule, where overlap: TERMINATE_OTHER supersedes a wedged run, but on the backfill path nothing supersedes anything, so a throttling Victor would retry a slice forever and hang the parent at Running. This bound is what converts that stall into a slice error the parent can report.

    A 7-day slice is at most 501 pages; at ~1s/page that is ~10 minutes, so an hour is ~6x headroom over the worst legitimate case. A workflow arg, so a slow-but-healthy Victor is retunable at run time rather than in code.