https://flyte.org logo
Join the conversationJoin Slack
Channels
announcements
ask-the-community
auth
conference-talks
contribute
databricks-integration
datahub-flyte
deployment
ecosystem-unionml
engineeringlabs
events
feature-discussions
flyte-bazel
flyte-build
flyte-console
flyte-deployment
flyte-documentation
flyte-github
flyte-ui-ux
flytekit
flytekit-java
flytelab
great-content
hacktoberfest-2022
helsing-flyte
in-flyte-conversations
introductions
jobs
konan-integration
linkedin-flyte
random
ray-integration
ray-on-flyte
release
scipy-2022-sprint
sig-large-models
workflow-building-ui-proj
writing-w-sfloris
Powered by Linen
ask-the-community
  • s

    sapcode

    11/06/2022, 7:50 AM
    Dear flyte team, we are about to POC an on-premise full(no sandbox) installation in a local kubernetes cluster. However in the documentation we cant find the sequence of which helm values yamls are supposed to be used when the target is not AWS and not GCP. It would be very helpful to have a section for non cloud installation setup without sandbox. 1) Can someone post an easy installation script for minikube or bare metal kubernetes, without external iDP provider? 2) Is there a build in user management, where we can create users ? 3) is there a default user and password? Some blogs are mentioning foobar. 4) it would be great to have 1 installation yaml for keycloak and the setup values documented which needs to be changed to get it running. Thank you in advance!!!
    f
    • 2
    • 13
  • f

    Felix Ruess

    11/07/2022, 1:30 PM
    Hi all, could the pod plugin be extended to allow using secrets as env vars and files? Right now it seems the env specified in pod spec is overwritten again... I cannot use flyte until I can create container tasks with some env vars and some files mounted from secrets.
    d
    e
    r
    • 4
    • 44
  • k

    Kamakshi Muthukrishnan

    11/07/2022, 2:17 PM
    #flyte-ray. Hi Looking for help to automatically adjust the memory size... Below is the error... int(100 * (memory / system_memory)))) ValueError: After taking into account object store and redis memory usage, the amount of memory on this node available for tasks and actors (-0.14 GB) is less than -13% of total. You can adjust these settings with ray.init(memory=<bytes>, object_store_memory=<bytes>).
    k
    • 2
    • 2
  • k

    Kamakshi Muthukrishnan

    11/07/2022, 3:58 PM
    @Ketan (kumare3) .Yes ray error when I run in flyte. this is a single node
    k
    d
    k
    • 4
    • 6
  • a

    Adedeji Ayinde

    11/07/2022, 6:47 PM
    Hi, is it possible to wrap a flyte task within a wrapper class implementation. When I tried calling the fit_classifier task from an object of the class I get an error message. Do you have any recommendation on how to wrap a flyte task in a class implementation?
    class model_wrapper:
        @task
        def fit_classifier(self, train: pd.DataFrame,
                           val: pd.DataFrame) -> JoblibSerializedFile:
    
            # fetch the features and target columns from the train dataset
            x = train[train.columns[1:]]
            y = train[train.columns[0]]
    
            # fetch the features and target columns from the validation dataset
            eval_x = val[val.columns[1:]]
            eval_y = val[val.columns[0]]
    
            m = self.xgb_classifier()
            # fit the model to the train data
            m.fit(x, y, eval_set=[(eval_x, eval_y)])
    
            working_dir = flytekit.current_context().working_directory
    
            fname = os.path.join(working_dir, "model.joblib.dat")
            joblib.dump(m, fname)
    
            # return the serialized model
            return JoblibSerializedFile(path=fname)
    Error message
    ...
    raise _user_exceptions.FlyteAssertion("Input was not specified for: {} of type {}".format(k, var.type))
    flytekit.exceptions.user.FlyteAssertion: Input was not specified for: self of type simple: NONE
    e
    n
    • 3
    • 7
  • r

    Rahul Mehta

    11/08/2022, 2:41 PM
    Has anyone noticed behavior where mypy doesn't catch invalid usages inside `@flytekit.task`/`@flytekit.dynamic`? We have a clear case w/ a missing positional argument that we'd typically expect the type-checker to catch, but it seems like the decorator may be causing some unintended behavior here?
    k
    e
    • 3
    • 11
  • a

    Andrew Achkar

    11/08/2022, 8:52 PM
    Is
    flytectl config init
    expected to respect a given
    --config
    flag to set the output location? AFAICT, it always writes to
    ~/.flyte/config.yaml
    e
    • 2
    • 3
  • v

    varsha Parthasarathy

    11/09/2022, 9:07 AM
    Hi team, Can you help with this cryptic error while registering a workflow
    raise _InactiveRpcError(state)
    grpc._channel._InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
    	status = StatusCode.RESOURCE_EXHAUSTED
    	details = "received initial metadata size exceeds limit"
    	debug_error_string = "{"created":"@1667984783.531503108","description":"Error received from peer ipv4:10.162.12.35:443","file":"src/core/lib/surface/call.cc","file_line":953,"grpc_message":"received initial metadata size exceeds limit","grpc_status":8}"
    s
    • 2
    • 11
  • m

    Mücahit

    11/09/2022, 12:29 PM
    Hey! Is there any guideline on Flyte and k8s version compatibility? I want to make sure k8s upgrades do not break Flyte components and Flyte helm chart(flyte-core)?
    s
    e
    • 3
    • 2
  • h

    Haytham Abuelfutuh

    11/09/2022, 3:16 PM
    @Akash
  • v

    varsha Parthasarathy

    11/09/2022, 7:40 PM
    👋 - I have a
    FlyteFile
    being passed as an input and my function (which i can’t edit) is expecting a os.path version of a file. How can I extract the
    uri
    from a
    Flytefile
    ?
    j
    • 2
    • 1
  • y

    Yee

    11/09/2022, 11:08 PM
    @Louis DiNatale you might also want to check out the k8s events
    l
    • 2
    • 5
  • f

    Frank Shen

    11/09/2022, 11:38 PM
    Hello, I have some python dependencies including custom packages to use in my flyte workflow. I want to first try them out on a local flyte cluster. How do I install them on the local flytesnack or demo cluster ? e.g. I have the dependencies scikit-learn, xgboost and some custom packages in a local wheel file I can do pip install -r requirements.txt --index-url https://maven.homebox.com/repository/max-pypi-snapshots/simple --extra-index-url https://maven.homebox.com/repository/max-pypi-releases/simple --use-pep517
    k
    • 2
    • 11
  • f

    Frank Shen

    11/10/2022, 12:05 AM
    Thank you in advance!
  • r

    Robin Eklund

    11/10/2022, 12:25 PM
    Hi! How do i set a default
    kubernetes_service_account
    for a specific launch plan (the value you can see in the picture)? i tried to do like this:
    LaunchPlan.get_or_create(
        name="my_lp",
        workflow=my_wf,
        auth_role=AuthRole(
            assumable_iam_role=None,
            kubernetes_service_account="my-flyte-service-account"
        ),
    )
    But when i am clicking "Launch Workflow", select the Launch Plan in the drop down, click "Advanced options" the values are empty (see picture). I also tried to do like this:
    LaunchPlan.get_or_create(
        name="my_lp",
        workflow=my_wf,
        security_context=security.SecurityContext(
            run_as=security.Identity(
                iam_role="default",
                k8s_service_account="my-flyte-service-account",
            ),
        ),
    )
    But this didn't work either. Anyone know how to do this?
    s
    • 2
    • 5
  • r

    Rahul Mehta

    11/10/2022, 2:19 PM
    Hey, I'm noticing that none of the logs we produce in our various tasks are actually surfaced when viewing pod logs (I only see the flytekit warnings about pickle types) -- is there something in particular we need to set to make logging work?
    k
    y
    • 3
    • 23
  • a

    Aphra Bloomfield

    11/10/2022, 4:38 PM
    Hey, all! First post here 🙂 Does there exist a more recent outline on automated deployment of Flyte to AWS (EKS)? I’m trying to follow this guide, but Opta seems to encounter YAML validation errors when following these steps.
    y
    t
    • 3
    • 8
  • a

    Alex Pozimenko

    11/11/2022, 7:26 PM
    hi, is there an option to set expiration date for intermediate (or even final) Flyte workflows outputs after which these artifacts will be permanently deleted from S3?
    k
    • 2
    • 1
  • f

    Frank Shen

    11/11/2022, 8:35 PM
    Hello, I have been trying to use the input params to the workflow but not able to due to they are Promises. Here is my test workflow to prove my point.
    @task
    def t1(team: str, a: int) -> Tuple[str, int]:
        print(f'in t1() team: {team}')
        print(f'in t1() a: {a}')
        return team, a
    
    @workflow
    def wf(team: str = 'abc', a: int = 3) -> Tuple[str, int]:
        print(f'in wf() team: {team}')
        print(f'in wf() a: {a}')
    
        x, y = t1(team=f'{team}_{a}', a=a)
        return x, y
    
    if __name__ == "__main__":
        print(f"result from wf(): {wf(team='abc', a=3)}")
  • f

    Frank Shen

    11/11/2022, 8:36 PM
    If I run it as a python program, I got:
    in wf() team: abc
    in wf() a: 3
    in t1() team: abc_3
    in t1() a: 3
    result from wf(): ('abc_3', 3)
    k
    g
    • 3
    • 8
  • f

    Frank Shen

    11/11/2022, 8:37 PM
    However, if I run it as a flyte workflow, I got:
    in wf() team: Promise(node:.team)
    in wf() a: Promise(node:.a)
    in wf() team: Resolved(team=scalar {
      primitive {
        string_value: "abc"
      }
    }
    )
    in wf() a: Resolved(a=scalar {
      primitive {
        integer: 3
      }
    }
    )
    in t1() team: Resolved(team=scalar {
      primitive {
        string_value: "abc"
      }
    }
    )_Resolved(a=scalar {
      primitive {
        integer: 3
      }
    }
    )
    in t1() a: 3
    result from wf(): DefaultNamedTupleOutput(o0='Resolved(team=scalar {\n  primitive {\n    string_value: "abc"\n  }\n}\n)_Resolved(a=scalar {\n  primitive {\n    integer: 3\n  }\n}\n)', o1=3)
  • f

    Frank Shen

    11/11/2022, 8:38 PM
    How do I get the real values of the input params in the workflow function to operate on them instead of passing them directly to the tasks?
    s
    • 2
    • 4
  • a

    Akash

    11/12/2022, 7:28 AM
    This error i m getting while running the python script in getting started part of Flyte
    k
    y
    • 3
    • 36
  • s

    seunggs

    11/12/2022, 6:00 PM
    Did anyone run into this issue? I’m running pip-compile with flytekit v1.2.3 in my requirements.in file and getting this error during the build:
  • s

    seunggs

    11/12/2022, 6:00 PM
    Could not find a version that matches protobuf<4,>=3.5.0,>=3.6.1,>=4.21.6 (from flytekit==1.2.3
  • s

    seunggs

    11/12/2022, 6:00 PM
    #15 13.29 There are incompatible versions in the resolved dependencies:
        #15 13.29   protobuf>=4.21.6 (from grpcio-status==1.50.0->flytekit==1.2.3->-r /root/requirements.in (line 1))
        #15 13.29   protobuf<4,>=3.6.1 (from flytekit==1.2.3->-r /root/requirements.in (line 1))
        #15 13.29   protobuf<4.0.0,>=3.5.0 (from flyteidl==1.1.22->flytekit==1.2.3->-r /root/requirements.in (line 1))
        #15 ERROR: executor failed running [/bin/sh -c pip-compile --output-file=/root/requirements.txt /root/requirements.in]: exit code: 2
  • s

    seunggs

    11/12/2022, 6:19 PM
    Looks like there’s a ticket for this - https://github.com/flyteorg/flyte/issues/3006. Temporarily pinning grpcio-status to <1.49.0 fixes the issue
  • t

    Tal

    11/13/2022, 3:52 PM
    Hi, A question - how can I manually trigger an upload (to s3 say) of a FlyteFile / FlyteDirectory?
    k
    y
    • 3
    • 24
  • z

    Zhiyi Li

    11/14/2022, 1:52 PM
    Does anybody ever encountered datacatalog pod hanging in the init container stage? It seems my pod stuck in the waiting status forever. Yet once I deleted the pod init container, the pod is up. Besides I’m wondering what does the datacatalog migrate actually do, would it be ok to skip this step if no db changes occurs?
    d
    k
    • 3
    • 24
  • l

    Laura Lin

    11/14/2022, 6:38 PM
    Are
    >>
    operators (to explicitly chain tasks) supported for dynamic workflows when running locally? either thru python or pyflyte run. It seems like its not being overridden. When I run it from the Flyte console, it doesn't throw any errors though.
    z >> x
    TypeError: unsupported operand type(s) for >>: 'str' and 'DataFrame'
    s
    y
    • 3
    • 13
Powered by Linen
Title
l

Laura Lin

11/14/2022, 6:38 PM
Are
>>
operators (to explicitly chain tasks) supported for dynamic workflows when running locally? either thru python or pyflyte run. It seems like its not being overridden. When I run it from the Flyte console, it doesn't throw any errors though.
z >> x
TypeError: unsupported operand type(s) for >>: 'str' and 'DataFrame'
s

Samhita Alla

11/14/2022, 6:46 PM
Should be supported. Could you share what
z
and
x
are?
l

Laura Lin

11/14/2022, 7:07 PM
from flytekit import task, workflow, dynamic


@task()
def test(s: str) -> str:
    return s


@task()
def test2(s: str) -> str:
    return s


@dynamic
def xf():
    z = test(s="yo")
    x = test2(s="test2")
    z >> x
    return x

@workflow
def wf():
    return xf()


if __name__ == "__main__":
    print(wf())
here's an example where I get
line 24, in xf
    z >> x
TypeError: unsupported operand type(s) for >>: 'str' and 'str'
y

Yee

11/14/2022, 9:17 PM
we are working on this actually. this is a known bug.
will have an issue written up and hopefully a fix later in the week.
i assume you’re seeing this in local execution?
l

Laura Lin

11/14/2022, 9:36 PM
yep.
@Yee is there a github issue I can follow?
y

Yee

11/14/2022, 9:37 PM
not yet
we’ll try to make one.
(we will make one… it’s just that it’s one of those where by the time you figure out how to write a nice ticket, you’ve already fixed it)
https://github.com/flyteorg/flytekit/pull/1343 - wip
hope to merge soon
View count: 2