Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Can you can the api and return all flow_runs with a specific Parameter?

you can check for whatever param values you want, and get more specific about where you look with the `flow_filter` and/or `flow_run_filter`
```In [1]: from prefect import get_client

In [2]: async with get_client() as client:
   ...:     runs = [run for run in await client.read_flow_runs() if run.parameters != {}]
   ...:

In [3]: runs```

<@U02GMEZU18B> I have a bunch of pending jobs clogging my queue where some worker failed.  How do i write a script similar to the one above to cancel all task flows in a pending state

I tried this
```    async with get_client() as client:
        runs = [run for run in await client.read_flow_runs() if run.state.is_pending()]
        print(runs)```

hmm - just a small sanity check, are you running the script from the same workspace as your pending runs? your client use looks fine

hmm could you grab an ID and/or filter on name of that seems pending and read that flow run to see what the API thinks it is?

like it might be LATE or something is what i am thinking

<@U02GMEZU18B> it is pending, but it looks like the issue is that `await client.read_flow_runs()` only pulls back 200 flows.  DO you have an example of how to pass a filter for pending ?

```In [15]: from prefect.client.schemas.filters import FlowRunFilter, FlowRunFilterState, FlowRunFilterStateType

In [16]: async with get_client() as client:
    ...:     runs = await client.read_flow_runs(
    ...:         flow_run_filter=FlowRunFilter(state=FlowRunFilterState(type=FlowRunFilterStateType(any_=["PENDING"])))
    ...:     )
    ...:```

Is it possible to remove the 200 limitation by just filtering by date range (_after _ before)?

I haven't found one, but you can use the offset field to write a loop until you run out of responses

yes, that is what I’ve done although it is a bit verbose code-wise but it works