Hi, I am trying to install the

metaflow-ui

helm chart from a local repo (cloned from your

metaflow-tools

repo), but my pod for the

backend-service

crashes... the issue is related to the

metadatadb

. Our metaflow service (not the ui) was deployed using your teraform templates, so we have the service in a EC2 load balancer and a database in RDS. I am not sure what host should I use in this section of the values.yaml for the metaflow-ui chart

metadatadb:
  port: 5432
  password:
  host: 
  user:

Also, what should be the value for

METAFLOW_DATASTORE_SYSROOT_S3:

should it be the same value used in the

./metaflowconfig/config.yaml

?

METAFLOW_DATASTORE_SYSROOT_S3:

He Alexadner, just to confirm that I understand your issue correctly, you are are trying to deploy metaflow ui on k8s using the ui specific helm chart. You have already deployed the metadata-service on an EC2 instance and have an RDS instance that is being used as your metaflow db. I am assuming you are using an S3 bucket as your data storage. Based on this: 1. For DB, yes you should provide the RDS's user, password and host address (RDS DB's host name), and also make sure there is a network line of sight between your UI pod in k8s and the RDS instances. 2. Yes, the value for

METAFLOW_DATASTORE_SYSROOT_S3

should match your local one. Let me know if you have more questions.

Hi @hundreds-zebra-57629 thank you for your reply, sorry that I have been away... You are correct, that's exactly what I would like to do. I tried with your suggestions and the deployment is still failing. Maybe you can help me with these questions: 1. What should be the value here. In the chart it seems like we are pointing the ui to a local metaflow service, but that is not our case, right?. I have tried replacing that value with the value I have in my

.metaflowconfig/config.json

for `METAFLOW_SERVICE_URL`but still fails. 2. What do you mean with having a network line of sight between your UI pod in k8s and the RDS instances? 3. I get this from my pod: Liveness probe failed: Get "http://10.151.13.229:8083/api/ping": dial tcp 10.151.13.2298083 connect: connection refused

Hi @fast-pizza-24629 1. can you share the error you are seeing? 2. By network line of sight, what I mean is that there is routing setup between the k8s worker nodes and your RDS instance (this should be good if they are in the same VPC) and the security group on your RDS instance allows inbound traffic from the k8s worker nodes. 3. Sounds to me like the metadata service on the instance with ip

10.151.13.229

is not listening on port

8083

. This could be because it is the wrong port or the service is not configured to receive traffic from outside the node. Can you double check if the service is listening on

0.0.0.0:8083

in the instance

10.151.13.229

. BTW I may be wrongly assuming

10.151.13.229

is the IP of your EC2 instance. If you have some time tomorrow, I'd happy to get on a call any time during PDT working hours.

Hi @hundreds-zebra-57629 (sorry, I have been OOO all this time). It would be great to have the call (if the offer is still available). I am located in Copenhagen, but can pretty much meet at the time that fits you better. Please let me know 🙂

Hey @fast-pizza-24629 if it works for you, lets do 9AM PDT tomorrow (Aug 16th).

Sounds good, thank you!

Massive apologies!, I didn't see your message in time. Friday same time works for me.

No problem, see you on Friday. Thank you 🙏

@fast-pizza-24629 I am ready whenever you are

I am here

Yep Huddle works