Running pyfluent in GitHub workflow fails

[vgelbgras]

Hi guys

I am trying to enable end-to-end tests using pyfluent.
Locally the tests works.
Through a GitHub workflow on the same machine (self-hosted Windows 10), they failed with: failed to connect to all addresses
I tried the suggestion from from #1894, it did help.

Any suggestions to debug?

[raph-luc]

Can you provide more details? Do you have a link to these GitHub actions that are failing that we can look at?

[vgelbgras]

Here is a run
https://github.com/ansys-internal/glow-engine/actions/runs/9062484001

Errors are similar to: failed to connect to all addresses; last error: UNAVAILABLE: ipv4:127.0.0.1:56802: Connection refused
The port is not always the same

The tests are using pyfluent through glow-engine. Locally they takes around 40seconds each.

In the pyfluent log I see only

2024-05-13 14:26:44,088 pyfluent.networking   DEBUG    GRPC_TRACE: RPC = /grpc.health.v1.Health/Check, request = {}
2024-05-13 14:28:01,406 pyfluent.networking   DEBUG    GRPC_TRACE: RPC = /grpc.health.v1.Health/Check, request = {}
2024-05-13 14:29:24,201 pyfluent.networking   DEBUG    GRPC_TRACE: RPC = /grpc.health.v1.Health/Check, request = {}

[vgelbgras]

On linux (Ubuntu-20.04), the same tests are passing https://github.com/ansys-internal/glow-engine/actions/runs/9075177063/job/24935285128

[raph-luc]

FYI I have created an issue tracker for this: #2848