Update project-faq.md

nav_order: 3

---

- [Unexplained disconnections using an egress service](#faq-1)

- [FAQ 1 - Unexplained disconnections using an egress service](#faq-1-unexplained-disconnections-using-an-egress-service)

- [FAQ 2 - K8s Docker container runtime deprecation](#faq-2-k8s-docker-container-runtime-deprecation)

-----

## FAQ 1

## FAQ 1: Unexplained disconnections using an egress service

_**When using an egress service with an external node, unexplained session disconnections are observed**_

"Egress service" is an AdvantEDGE feature used to include [external edge nodes]({{site.baseurl}}{% link docs/overview/features/overview-external-nodes.md %}) into a scenario. On certain systems, configuring & using an egress service will present unexplained spurious disconnections.

       Kernel modules: e1000e

```

To temporarily disable segmentation offload

To temporarily disable segmentation offload:

```

# xxx is the device name (eth0, eno1... etc)

ethtool -K xxx gso off gro off tso off

```

To permanently disable segmentation offload, modify the interface configuration file (`/etc/network/interfaces`) as shown below

To permanently disable segmentation offload, modify the interface configuration file (`/etc/network/interfaces`) as shown below:

```

iface eno1 inet static

address xx.xx.xx.xx

Related resources:

- https://serverfault.com/questions/193114/linux-e1000e-intel-networking-driver-problems-galore-where-do-i-start

- https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1766377

- [Serverfault question thread](https://serverfault.com/questions/193114/linux-e1000e-intel-networking-driver-problems-galore-where-do-i-start)

- [Ubuntu bug report](https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1766377)

-----

## FAQ 2: K8s Docker container runtime deprecation

_**Starting with k8s version 1.22, docker container runtime is no longer supported**_

With release 1.20, Kubernetes announced the deprecation of Docker as a container runtime, promoting instead other runtimes that support the

Container Runtime Interface (CRI) such as containerd and CRI-O. As of releses 1.22, Docker container runtime was officially unsupported.

**Why deprecate Docker container runtime?**

To support interoperability with different container runtimes, Kubernetes requires a container runtime that implements the Container Runtime Interface (CRI).

Docker container runtime is not CRI-compliant and requires an adaptation layer called _dockershim_ maintained by k8s. As of release 1.22, k8s decided to

no longer maintains _dockershim_ for Docker container runtime, in favor of CRI-compatible runtimes.

More details about the deprecation can be found here:

- [K8s dockershim deprecation v1.20 release notes](https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG/CHANGELOG-1.20.md#dockershim-deprecation)

- [K8s dockershim deprecation FAQ](https://kubernetes.io/blog/2020/12/02/dockershim-faq/)

- [K8s & Docker](https://kubernetes.io/blog/2020/12/02/dont-panic-kubernetes-and-docker/)

- [K8s Docker deprecation explanation](https://medium.com/better-programming/kubernetes-is-deprecating-docker-8a9f7566fbca)

**Impact on AdvantEDGE deployment**

AdvantEDGE [runtime installation procedure](https://interdigitalinc.github.io/AdvantEDGE/docs/setup/env-runtime/) deploys Docker as the

k8s container runtime. For this reason, AdvantEDGE installation as currently documented only supports k8s versions up to 1.21. We have validated

that the Docker container runtime can be replaced seamlessly by _containerd_ for k8s version 1.20, however we have not yet tested with the latest

k8s release 1.22.

We will update the AdvantEDGE documentation with the new container runtime installation procedure as soon as we officially support 

the latest k8s version. Users wishing to migrate to another CRI-compatible container runtime may do so at any time.