Hey team, is there a way to tell prefect “Retry all failed flows in the last day/hour/etc”?

Hi @Samuel Hinton - there's no direct call you could make to do this but you could use the client by querying for all failed flows after a certain time period, so something like this:

import datetime
from prefect import Client
client = Client()
today = datetime.date.today()
yesterday = today - datetime.timedelta(days=1)

query = f"""
query {
  flow_run(where: {state: {_eq: "Failed", scheduled_start_time: { _gte: {yesterday} } }}) {
    id
    state
    task_runs {
      id
    }
  }
}
"""

result = client.graphql(query)

# iterate through results to set failed task runs to "Pending"
# and re-schedule the flow run

That looks like a great solution. As someone unfamliar with graphql, are there examples or documentation on the prefect site as to how to reschedule the results of that query?

We don't have any docs on this particular use case, but you'd use a mutation like this:

mutation setTaskRunStates($input: [set_task_run_state_input!]!) {
  set_task_run_states(input: { states: $input }) {
    states {
      id
    }
  }
}

to set task run states, and:

mutation SetFlowRunStates($flowRunId: UUID!, $version: Int!, $state: JSON!) {
  set_flow_run_states(
    input: {
      states: [{ flow_run_id: $flowRunId, state: $state, version: $version }]
    }
  ) {
    states {
      id
      status
      message
    }
  }
}

to set flow run states; you'd want to make sure to set the task run states before setting the flow run state to

Scheduled

. For more help using graphql, check out the graphql and hasura docs.