[Azure] Deployment failed with "Virtual network site is not valid"

[Po-Hsu Lin]

Happened with 4.15 kernel on Xenial with the unionmount-overlayfs test, the deployment does not work on any node:

ERROR: Deployment failed. Correlation ID: 5dc3794d-f239-4991-8de0-ea213ca58082. {

  "error": {

    "code": "NetcfgInvalidVirtualNetworkSite",

    "message": "Virtual network site is not valid as specified with the name 'x-la-edge-azure-edge-4-15-0-Basic-A1-unionmount-overlayfs-suiteVNET'.",

    "details": []

  }

Retried for many times but still no luck.

[Po-Hsu Lin (cypressyew)]

Should be able to close this now.

[Sean Feole (sfeole)]

This is still an issue and an annoying one at that, What I believe is happening is that there is a character limit for resource group titles. and we are clearly hitting this limitation on -edge kernels because the names are so long. What we are not realizing is that there is additional characters that WILL BE APPENDED to this title. These characters are related to azure and range from '-networkIDXXXXXX' to '-subnetXXXX' (This all happens behind the scenes on azure network) So this makes sense why these were failing to begin with.

this only happens with -EDGE kernels because we have an additional 22 characters in each -edge-title.

This needs to be fixed asap and will add it to my list of things to do.

[Po-Hsu Lin (cypressyew)]

I think maybe this should be fixed in the az command, the tool should have the capability to catch exceptions and tell use this name is too long.

Found this article, which the tool should follow.
https://docs.microsoft.com/en-us/azure/architecture/best-practices/naming-conventions

[Sean Feole (sfeole)]

pushed back to top of priority q

[Sean Feole (sfeole)]

This was fixed in ckct 2 weeks ago. Marking fixed release as I have not seen this failure since.