[BUG-NOK] Open-Edge-Node - Inconsistent Auto Cluster Creation

[VishnuPriyaKasula]

Bug Description

In Edge Orchestrator v3.1, auto cluster creation is inconsistent.

• In some cases, the cluster is created automatically as expected.
• In other cases, the cluster is not created, leaving the provisioned host without a cluster.
  This issue is reproducible in both on-prem and AWS deployments.

System Setup

• Orchestrator Version: 3.1
• Deployments: On-prem and AWS

Reproducible Steps

1. Log in to Edge Orchestrator v3.1.
2. Navigate to Infrastructure → Hosts → Register Hosts.
3. Enter Host Name and Serial Number.
4. Enable the following options:
5. Onboard Automatically
6. Provision Automatically
7. Create Single-host Clusters
8. Save and allow the host to complete onboarding and provisioning.
9. Go to Hosts → Provisioned.
10. Check if the cluster is automatically created.

[krishnajs]

@VishnuPriyaKasula can you please share the screenshot of the clusters tab. Also, the edge node is in rebooting state. Did you see this after EN came aback online ?

[adorney99]

@VishnuPriyaKasula I've tried to reproduce a few times but haven't hit any problems. This looks like it could be an issue with your edge node infrastructure itself since you see a rebooting issue. Here are a few things to check:

• provisioning logs
• node agent logs
• cluster agent logs
  All of these can be accessed through the observability dashboard. Depending on how far provisioning/cluster creation progressed you'll only have access to some of these logs.

Something else to check would be the logs of the cluster-manager pod on your orchestrator cluster.
If it's not an infrastructure issue it may be something UI related. If logs all look fine something to check would be network traffic in the chrome console to see if requests get sent to cluster manager

[VishnuPriyaKasula]

The device came to running state , I just captured the device while it is in reboot state

[krishnajs]

@VishnuPriyaKasula great ! Can we close the ticket ? If you see this issue again, let's capture the logs and re-open if you see the issue.

[amathan]

@krishnajs the screenshot while creating the bug was taken when node was rebooting, but later @VishnuPriyaKasula has provided when device came online, but cluster auto creation is still a problem.

[krishnajs]

Thanks @amathan for the update. Can you share the screenshot of the cluster page and logs suggested by @adorney99 #746 (comment)

[amathan]

We re-tested with two devices, one cluster was auto created, and another dint create.

Note: And another bug #768 is created where we see provision is struck at Rebooting not sure it is interlinked to cluster creation logic

[adorney99]

@amathan can you try gathering the logs listed above to see if they have any info on what might be causing this issue? I see one node is in an error state. Is that the one that didn't create a cluster? If so it could be an issue with the edge node rather than cluster creation

[amathan]

@adorney99 The error state is a usual observation when agents are getting installed, you can see below the state changed to provisioned from Error but still no cluster created for it - Device PF1236KN.

[adorney99]

@amathan sorry didn't see your response. When the host goes to provisioned are you able to create a cluster with it manually? If this is an issue with cluster creation you'll need to gather the logs outlined here so we can try to figure out what's going wrong during provisioning/cluster creation