hello, where can I download the images for flyte-sandbox

like 'cr.flyte.org/flyteorg/flytescheduler-release:v0.19.3'

sorry I have a problem with installing flyte-sandbox from a command 'flytectl sandbox start', I must download all docker images locally, and install them.

:flyte: It’s time for our weekly Office 🏢 Hours ⌛! This week I’ll be on the zoom link for the next 30 mins. Zoom link

Hi Flyers, Watch the scoop on Flyte's spot instances and interruptible tasks, checkpointing, and the all-new Flyte Decks by Union.ai's @Ketan (kumare3). Also catch this week's . Meeting minutes: https://bit.ly/3NdCL2h See you on April 5th!

Its time for our weekly office hours 🏢 - :flyte: . I will be on the Zoom link in 10 minutes and stay there for 30 minutes until 9:30 PST

📣 Contributors of the Month 📣 Hello everyone! With the inflow of contributions, Flyte is being constantly improved. On that note, we want to give shout-outs and send swag to all our contributors who are sweating it out to make Flyte even better! The shout-outs are going to happen every month. The following is the list of contributors we’d like to thank this month. • @Sören Brunk has been extensively working on simplifying the deployment of Flyte on GCP, among other backend improvements. He’s one of the core contributors! • @Max Smolin worked on a significant PR that added support for Union type, one of the most requested features. • @Kenny Workman worked on adding type annotation support to the Flytekit codebase. We hope to have many more of your contributions, Soren, Max, and Kenny! All contributors of the month should be available on flyte.org.

Please clap on this post by Lyft - pretty awesome!!!! = https://eng.lyft.com/orchestrating-data-pipelines-at-lyft-comparing-flyte-and-airflow-72c40d143aad

If I want to deploy flyte-sandbox to AWS, what is the minimum EC2 that I need?

Hi, I want to deploy Flyte Helm charts to our EKS cluster. We currently have an older version of AWS Ingress Controller than required for the Ingress. Is it possible to bypass this requirement by e.g., port-forwarding and still being able to deploy workflows to

FlyteAdmin

? It is in our road map to bump the ALB Ingress Controller version, but until that is done I would still like to experiment with Flyte on EKS.

has anyone successfully benchmarked propeller for >5k concurrent workflows? what does the propeller config look like for something like that?

Hi! I was to access flyteconsole service externalize endpoint. I have followed the complete document: https://docs.flyte.org/en/latest/deployment/gcp/manual.html#deployment-gcp-manual The workload and service are up (the screenshot is shared): • The docment says --> Use the https://<FLYTE-ENDPOINT>/console to get access to flyteconsole UI I have done the same and used the domain as flyte-endpoint but nothing is showing up. Are there any further steps to access flyteconsole UI ?? like create external IP instead of cluster IP or mapping service to domain

Hi I am doing

helm install incubator/sparkoperator --namespace spark-operator --kubeconfig=~/.flyte/k3s/k3s.yaml

Error: Error: INSTALLATION FAILED: must either provide a name or specify --generate-name What is generate name here?

Hey Flyte Team, I just updated to Flyte

v0.19.3

. No matter what I add to my Propeller

deployment.yaml

, I cannot seem to get Stackdriver log links to render in Flyte Console for map tasks (i.e.

k8s_array

tasks). Map task subtasks always render with Kubernetes log links (even when declaring

kubernetes-enabled: false

). I'm pretty sure that Stackdriver log links just don't work yet for map tasks. Could you confirm that?

@Yuvraj Propeller workload pod is still crashing. Annotation with a namespace is done. Also, the actual problem statement is still there. As I need to access external IP for UI services. @Yee

hi folks, i'm looking at available Flyte metrics to help us monitor the platform as we run large workloads (and stress test the cluster). Are there good docs/samples available? I found this but it's very high level. One specific question: Is this metric cumulative -

flyte-propeller-all-workflow-acceptance.latency.ms.sum

? Sample chat below. Or perhaps I should be looking at percentiles only (.ms.0.9 and .ms.0.99)?

Hi! We added a timeout for one node in our workflow. This node is handled by a backend plugin (

bigqueryjob

). We see in the logs that the node timeouts as expected:

"Current execution for the node timed out; timeout configured: 3h0m0s"

However, the

Delete

method of the plugin (https://github.com/flyteorg/flyteplugins/blob/master/go/tasks/plugins/webapi/bigquery/plugin.go#L226) is never called after the timeout, so the job is not cancelled. Is this expected? How can I tell propeller to abort the node after a timeout?

Hi, I have deployed Flyte on EKS and I am experimenting with the fast register approach as I am only iterating my source code. I saw that I should configure flytectl with S3 configuration. It works with “non-fast” registration but I am getting this error with fast registration. I have serialized using the fast approach initially. Any ideas what might cause this error?

(.venv) ☁  ~ ⚡ make register PROJECT=determined DOMAIN=development
flytectl register files \
		 --project determined \
		 --domain development \
		 --archive flyte-package.tgz \
		 --version v1-fast1
Error: please check your Storage Config. It failed while uploading the source code. Failed to write data [69b] to path [fast/v1-fast1-fast9bb248f6c3ae89386ec7105633eaaee8.tar.gz].: PutObject, putting object: NoCredentialProviders: no valid providers in chain. Deprecated.
	For verbose messaging see aws.Config.CredentialsChainVerboseErrors
gmake: *** [Makefile:22: register] Error 1

(.venv) ☁  ~  cat ~/.flyte/config.yaml
admin:
  endpoint: dns:///localhost:8089
  authType: Pkce
  insecure: true
logger:
  show-source: true
  level: 0
storage:
  type: stow	
  stow:
    kind: s3
    config:
      auth_type: iam
      region: eu-west-1
  container: <name-of-my-bucket>

hello there 🙂

i’m really struggling with getting flyte work on k3d , also the docker-desktop kubernetes , are the docs up to date ? anyone else tried to have it deployed on k3d. ?

Hey, everyone! I'm trying to integrate our MLflow server to our Flyte projects. My current approach consists of passing the MLflow username and password as Secrets to the required tasks, but I was wondering: does anyone else here uses this stack, and if so, are you willing to share a little bit about how it's configured? 🙂

There’s a new use case that came up and we’re trying to figure how best to do this. I was wondering if I could some help with resources/ideas. So we have a bunch of tasks that are already registered and at runtime we want to be able to construct the graph. One thought was to basically create the workflow graph spec and submit a request to flyteadmin similar to how flytekit does. Then the newly created workflow can be launched. In short we want to be able to generate a workflow without needing to write python code for it. Just specify the proto spec. Thoughts? @katrina @Ketan (kumare3) @Yee cc: @Rupsha Chaudhuri @Danielle Curammeng

Are there any examples of flyte scheduling workloads on CoiledCloud ?

Hi Flyers.

flyte-admin-execution.manager-active.executions.gauge

shows negative values. Is this by design?

Hi Flyers, Save the date! Help us welcome @Jake Neyer from Striveworks, who will be presenting at our upcoming Flyte Community Sync next week. Details coming... Tuesday April 5, 9am PT

Hey folks, trying to debug an issue. We’re seeing various tasks fail with

terminated in the background, manually

. I took a look at the code here and see that this occurs when the pod get deleted. I don’t see a delete request being sent to admin so don’t think there is anything there. Propeller logs haven’t shown anything either. So I’ve been looking at kube logs to try to see what is triggering the pod deleted but there’s a lot of noise in the logs so taking some time to through them. Meanwhile, I figured I’d ask here if anyone has run into something similar before

hIts time for the weekly office hours. I am on the Zoom link, for the next 30 minutes. Drop in, say hi!

I need some help getting launchplans with a cronjob configured to work: I first register a launchplan with a

* * * * *

cron expression via flyteremote. After that, the flyte UI still shows

This workflow has no schedules

, as expected. I then execute

flytectl update launchplan -p default -d development --version fabio-cron-2a3fb7f-dirty-BQN44 cron_hydra_workflow_cfg_flyte_workflows.collaborative_filtering.workflow_0 --activate

which gives:

updated launchplan successfully on cron_hydra_workflow_cfg_flyte_workflows.collaborative_filtering.workflow_0%

In the UI, the schedule is updated as well, it now says

Every minute

. Problem: I’m not seeing any executions. The flyte admin shows this in the logs after I activated the launchplan:

{"json":{"src":"launch_plan.go:108"},"level":"info","msg":"Adding resource type for unspecified value in request: [%+v]id:\u003cproject:\"default\" domain:\"development\" name:\"cron_hydra_workflow_cfg_flyte_workflows.collaborative_filtering.workflow_0\" version:\"fabio-cron-2a3fb7f-dirty-BQN44\" \u003e state:ACTIVE ","ts":"2022-03-31T18:13:01Z"}
{"json":{"src":"log.go:42"},"level":"warning","msg":"Failed to parse authenticated client metadata when creating audit log","ts":"2022-03-31T18:13:01Z"}
{"json":{"src":"log.go:91"},"level":"info","msg":"Recording request: [{\"Principal\":{\"Subject\":\"\",\"ClientID\":\"\",\"TokenIssuedAt\":\"0001-01-01T00:00:00Z\"},\"Client\":{\"ClientIP\":\"\"},\"Request\":{\"Method\":\"UpdateLaunchPlan\",\"Parameters\":{\"domain\":\"development\",\"name\":\"cron_hydra_workflow_cfg_flyte_workflows.collaborative_filtering.workflow_0\",\"project\":\"default\",\"version\":\"fabio-cron-2a3fb7f-dirty-BQN44\"},\"Mode\":1,\"ReceivedAt\":\"2022-03-31T18:13:01.578706629Z\"},\"Response\":{\"ResponseCode\":\"OK\",\"SentAt\":\"2022-03-31T18:13:01.597983728Z\"}}]","ts":"2022-03-31T18:13:01Z"}

The

flytescheduler

doesn’t show any update in the logs. Does anyone have some guidance where to start debugging this? Thanks!

Hey Flyte Team, we run a telegraf sidecar container in our flyteadmin and flytepropeller pods to scrape Prometheus metrics (from flyteadmin and flytepropeller). Does flyteconsole support this as well or not yet?