apache / apache/cloudstack

VMware NSX network implementation can fail before backing DV port group is visible

Open
#13,752 2 comments 0 reactions 0 assignees Claimed by @Dogface2k View on GitHub
component:vmware component:vpc Severity:Major type:bug
Dominant language
Java
Stars
3.1k
Forks
1.4k
Avg merge
6d 19h
Merged PRs (30d)
32

Description

### Problem

On VMware environments using the NSX integration, implementing a guest network can fail during `PlugNicCommand` even though `CreateNsxSegmentCommand` has already succeeded.

The NSX segment is created outside vCenter. Its backing distributed virtual port group can therefore take a short time to become visible through the vCenter API. `HypervisorHostHelper.prepareNetwork` currently performs only one immediate lookup for NSX networks and returns failure when that first lookup is empty.

Sanitized management-server sequence:

```text
CreateNsxSegmentCommand ... NsxAnswer result=true
Prepare network on vmwaredvs
Failed to create guest network
PlugNicAnswer result=false
```

The failure rolls back the NIC attachment and prevents the persistent VPC guest network from being implemented.

### Versions

- Apache CloudStack 4.22.1.0
- VMware vSphere / vCenter 8.0.3
- CloudStack NSX integration

### Steps to reproduce

1. Configure a VMware zone with the CloudStack NSX integration.
2. Create an NSX-backed VPC guest network.
3. Allow the NSX segment command to complete while the backing DV port group is not yet visible through vCenter.
4. Observe the immediate DV port group lookup fail in `HypervisorHostHelper.prepareNetwork`, followed by a failed `PlugNicCommand`.

### Expected behavior

After successful NSX segment creation, CloudStack should use its existing bounded DV port group readiness wait before attempting to attach the NIC. If the port group never becomes visible, the existing timeout should still fail the operation cleanly.

### Proposed fix

Enable the existing `waitForDvPortGroupReady` path for NSX broadcast domains and add a regression test covering an empty first vCenter lookup followed by a successful lookup.

Contributor guide

Open the contributing guide

Research direction

Start in HypervisorHostHelper.prepareNetwork and trace the existing waitForDvPortGroupReady path used during PlugNicCommand. Add the regression test described in the issue for an empty first vCenter lookup followed by a successful lookup, and verify the bounded timeout still handles a port group that never appears.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
infrastructure, networking
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.