https://prefect.io logo
Join Slack
Powered by
Hey folks, I'm encountering an issue trying to get...
# ask-community
c
Hey folks, I'm encountering an issue trying to get a flow to rerun itself when a condition is met. Here's the code for my flow:
Copy code
rerun_flow = StartFlowRun(
        flow_name="blast_import_flow",
        project_name="python_imports"
        )   
spawn_time_series_import = StartFlowRun(
        project_name="python_imports", 
        flow_name="blast_metric_series_flow",
        wait=True)

with Flow("blast_import_flow", 
        schedule=daily_schedule, 
        state_handlers=[cloud_only_slack_handler]
       ) as flow:

    data_object = get_blast_batch()
    check_api_result = check_api(data_object=data_object)
    transformed_data_objects = transform(data_object=check_api_result)
    imported = import_to_database(data_object_list=transformed_data_objects)
    blast_ids, record_length = push_records_to_summary_table(imported)

    # package ids into the right dict format
    packaged_ids = package_into_params.map(blast_id=blast_ids)

    spawn = spawn_time_series_import.map(
            parameters=packaged_ids)
    with case(record_length, 5): 
        rerun_flow()
The ETL portions of the flow are working just fine, and the 
spawn_time_series_import
is correctly creating its tasks. But 
rerun_flow
isn't - I'm seeing this error:
Copy code
Unexpected error: ClientError([{'path': ['user'], 'message': 'field "user" not found in type: \'query_root\'', 'extensions': {'path': '$.selectionSet.user', 'code': 'validation-failed', 'exception': {'message': 'field "user" not found in type: \'query_root\''}}}])
Traceback (most recent call last):
  File "/usr/local/lib/python3.7/site-packages/prefect/engine/runner.py", line 48, in inner
    new_state = method(self, state, *args, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/prefect/engine/task_runner.py", line 860, in get_task_run_state
    logger=self.logger,
  File "/usr/local/lib/python3.7/site-packages/prefect/utilities/executors.py", line 298, in run_task_with_timeout
    return task.run(*args, **kwargs)  # type: ignore
  File "/usr/local/lib/python3.7/site-packages/prefect/utilities/tasks.py", line 449, in method
    return run_method(self, *args, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/prefect/tasks/prefect/flow_run.py", line 172, in run
    run_link = client.get_cloud_url("flow-run", flow_run_id)
  File "/usr/local/lib/python3.7/site-packages/prefect/client/client.py", line 887, in get_cloud_url
    tenant_slug = self.get_default_tenant_slug(as_user=as_user and using_cloud_api)
  File "/usr/local/lib/python3.7/site-packages/prefect/client/client.py", line 920, in get_default_tenant_slug
    res = self.graphql(query)
  File "/usr/local/lib/python3.7/site-packages/prefect/client/client.py", line 318, in graphql
    raise ClientError(result["errors"])
prefect.utilities.exceptions.ClientError: [{'path': ['user'], 'message': 'field "user" not found in type: \'query_root\'', 'extensions': {'path': '$.selectionSet.user', 'code': 'validation-failed', 'exception': {'message': 'field "user" not found in type: \'query_root\''}}}]
I'm not sure what the difference is, aside from the wait. I figure I don't want a bunch of these telescoping and waiting on each other to finish.
z
Hi @Chris Jordan -- are you using Prefect Server?
c
I'm using prefect cloud, registering these with an agent on a server of mine and running from the web gui
z
Could you give me the output of 
prefect diagnostics
locally and from the agent's machine?
c
locally here being the machine I'm developing on?
z
Yep!
c
local:
Copy code
(python-imports-3.7.6) ┌─[cjordan@WMNAPA-BBO02-20] - [~/work/python_imports] - [2021-02-04 05:56:36]
└─[0] <git:(final_blast bad0de3✈) > prefect diagnostics 
{
  "config_overrides": {
    "cloud": {
      "graphql": true
    },
    "context": {
      "secrets": false
    },
    "debug": true,
    "engine": {
      "executor": {
        "dask": {
          "address": true,
          "local_processes": true
        }
      }
    },
    "flows": {
      "defaults": {
        "storage": {
          "default_class": true
        }
      }
    }
  },
  "env_vars": [
    "PREFECT__CLOUD__AGENT__AUTH_TOKEN"
  ],
  "system_information": {
    "platform": "Darwin-19.6.0-x86_64-i386-64bit",
    "prefect_backend": "server",
    "prefect_version": "0.13.18",
    "python_version": "3.7.6"
  }
}
(python-imports-3.7.6) ┌─[cjordan@WMNAPA-BBO02-20] - [~/work/python_imports] - [2021-02-04 05:56:41]
└─[0] <git:(final_blast bad0de3✈) >
z
So your backend there is Server, not Cloud and you've overridden the API url?
c
yes the backend there is Server, but I'm only developing there
for deployment I push to the server that I'm running agent on
and register there
z
I see 👍
c
here's the diagnostics from that server
Copy code
cjordan@livo:/data/cjordan/python_imports$ prefect diagnostics 
{
  "config_overrides": {},
  "env_vars": [],
  "system_information": {
    "platform": "Linux-4.4.0-1100-aws-x86_64-with-debian-stretch-sid",
    "prefect_backend": "cloud",
    "prefect_version": "0.13.18",
    "python_version": "3.7.6"
  }
}
z
Great thanks! If you're willing to try the newest version of Prefect that'd be helpful, I'll be looking into some sort of breaking change here.
c
that seems plausible. I think there was a reason I didn't upgrade, though I can't recall it now, and I don't want to spend Thursday evening chasing that down
could you talk me through a guess of what's going on, and I can try to debug things a little myself?
z
It's failing here:
Copy code
def get_default_tenant_slug(self, as_user: bool = True) -> str:
        """
        Get the default tenant slug for the currently authenticated user

        Args:
            - as_user (bool, optional): whether this query is being made from a USER scoped token;
                defaults to `True`. Only used internally for queries made from RUNNERs

        Returns:
            - str: the slug of the current default tenant for this user
        """
        if as_user:
            query = {
                "query": {"user": {"default_membership": {"tenant": "slug"}}}
            }  # type: dict
        else:
            query = {"query": {"tenant": {"slug"}}}

        res = self.graphql(query)

        if as_user:
            user = res.get("data").user[0]
            slug = user.default_membership.tenant.slug
        else:
            slug = res.get("data").tenant[0].slug
        return slug
c
if my only option is to upgrade then I'm willing to work through that, but if I can skate by without for now I'd prefer that
z
Upgrading would just help nail down if it's a versioning issue
So the error is that there isn't a "user" field which generally indicates that the database schema isn't matching what the client expects
Usually this is a Prefect Server / Core version mismatch but we generally are careful not to break things like this with Cloud
c
got it, that makes sense
z
Ah hey this may be a token issue
c
I can take a look at upgrading, but the soonest I can devote nontrivial time to it is Tuesday
z
What scope of token are you using?
RUNNER / USER ?
c
one sec looking this up...
z
Which agent type are you using as well?
c
the token I'm using is at least a few versions old, and the note I saved it in only calls it a 
prefect agent token
looks like I'm starting a docker agent
invocation is here
Copy code
prefect agent start \
        --volume "${WORKDIR}/.config.toml:/opt/prefect/config.toml" \
        -t "${token}" \
        docker \
        --show_flow_logs=True
z
Could you try generating a new USER scoped token and using that instead?
c
I'll give that a shot, give me a minute
(lastpass is annoying to sync)
also let me double check here, cause the woding isn't consistent across all the documentation - from https://docs.prefect.io/orchestration/concepts/api.html#authentication, 
USER-scoped API tokens function as personal access tokens.
You seem to use the words interchangably
these are the same thing right?
z
Yeah this is in the middle of a refactor because it's really confusing as is... one sec
c
They're replacing a bridge next to my workplace, and for some time, every time I drive over the old one I think to myself "man this bridge is in really bad shape they should really do something about it" .... as the construction crews are working 30ft away
z
😄
What you want is a token that's not RUNNER scoped 😄
It turns out this is a known issue that was fixed in December (https://github.com/PrefectHQ/prefect/commit/810faa5039422ac2ae5e26de29902fd424c6f2b0)
Since we're not passing 
as_user=False
it's trying to access something that RUNNER tokens don't have permissions to see.
c
Copy code
(python-imports-3.7.6) cjordan@livo:/data/cjordan/python_imports$ screen -r 

    return self.main(*args, **kwargs)                                                           File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/core.py", line 782, in main                                                       rv = self.invoke(ctx)                                                                       File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/core.py", line 1259, in invoke                                                    return _process_result(sub_ctx.command.invoke(sub_ctx))                                     File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/core.py", line 1259, in invoke                                                    return _process_result(sub_ctx.command.invoke(sub_ctx))                                     File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/core.py", line 1066, in invoke                                                    return ctx.invoke(self.callback, **ctx.params)                                              File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/core.py", line 610, in invoke                                                     return callback(*args, **kwargs)                                                            File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/click/decorators.py", line 21, in new_func
    return f(get_current_context(), *args, **kwargs)
  File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/prefect/cli/agent.py", line 666, in start
    docker_interface=not no_docker_interface,
  File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/prefect/agent/agent.py", line 219, in start
    self._verify_token(self.client.get_auth_token())
  File "/home/users/cjordan/.pyenv/versions/3.7.6/envs/python-imports-3.7.6/lib/python3.7/site-packages/prefect/agent/agent.py", line 178, in _verify_token
    raise AuthorizationError("Provided token does not have a RUNNER scope.")
prefect.utilities.exceptions.AuthorizationError: Provided token does not have a RUNNER scope.
cjordan@livo:/data/cjordan/python_imports$
so it seems like a version upgrade is probably in order
Like I said I can't get to that for a couple of days
z
Aghh. I shall ask someone that understands our tokens. I think you just need to grab a TENANT token using your PAT/USER token
c
got it, thanks
this isn't a pressing issue - if someone gets back to me sometime Monday that would be fantastic
and in the meantime I can spend some time Tuesday working through upgrades
z
You can make a tenant token direectly at https://cloud.prefect.io/team/tokens
c
lemme see if I've got one already
looks like no
one sec, pushing it through the machine...
I'm getting the same runner error. I double checked that the token I just generated has the scope TENANT and that the token my script is pulling out of lastpass is the one that I just generated
er the error is
Provided token does not have a RUNNER scope.
z
Annoying, sorry about that.
It looks like it won't let you use a higher privilege token.
I think you'll either need to patch that line yourself or upgrade -- sorry it's broken on older versions 😕
c
can't fault you for not having fixed it fast enough for me to get the upgrade last time I did an upgrade
but this has been really helpful
z
👍 0.14 has all sorts of fun stuff in it
c
I'm glad to know that there is a fix, and I'll have the time to apply it soon
39 Views
Open in Slack
PreviousNext
Powered by