[Murano] Sometimes environment deployment failures due to Error: The murano-agent did not respond within 3600 seconds

[Oleh Baran]

Detailed bug description:
Deployment of Docker applications fails randomly with a message murano-agent did not respond within 3600 seconds. In log I observe ERROR muranoagent.app error: [Errno 32] Broken pipe, which always happens within exactly 2 hours after deployment starts and connect renews after that.

Steps to reproduce:
Deploy environment with Docker Crate, DockerNginX, DockerMongoDB applications in one Kubernetes container (Kubernetes cluster has: initial number of Kubernetes nodes: 1, max number of Kubernetes nodes: 2, initial number of gateway nodes: 1, max number of gateway nodes: 2, assign floating IP to Kubernetes nodes: true, assign floating IP to gateway nodes: true, expose cAdvisor UI: true. Kubernetes pod has number of replicas: 1)

Expected results:
Deployment should be successful

Actual result:
Deployment finishes with Error: The murano-agent did not respond within 3600 seconds

Reproducibility:
Randomly

Description of the environment:
ENVIRONMENT: MOS 9.0 ISO 305

Upstream bug: https://bugs.launchpad.net/murano-apps/+bug/1583007

[Stan Lagun (slagun)]

This is likely to be fixed already. Need to rebuild the image so that newer murano agent be used and try to reproduce the bug

[Timur Nurlygayanov (tnurlygayanov)]

Please try this image with the latest version of Murano agent:
https://www.dropbox.com/s/gwfkagz4o923bk0/ubuntu14.04-x64-agent.qcow2?dl=0

[Timur Nurlygayanov (tnurlygayanov)]

We tried new images with tha latest version of murano agent, the problem reproduced again.

[Serg Melikyan (smelikyan)]

Fix proposed in upstream by https://review.openstack.org/317127

[Serg Melikyan (smelikyan)]

we are waiting for the feedback from QA team regarding results

[Oleh Baran (obaran)]

I tried to deploy environments with fix https://review.openstack.org/#/c/317127/ and bug didn't reproduce. Seems like fix works.

[Victor Ryzhenkin (vryzhenkin)]

This bug in murano-agent component. So, finally, this one doesn't affects codebase and only affects murano-agent deliverables(images with m-agent in case).
So, according that, we should mark this bug as won't fix in MOS.