This is a common error for me and have been working closely with support on this very issue this for many many months. Escalated through support in various channels right up to the top. Others are also having similar issues as you can see here: umbraco cloud manually trigger build and deploy without git check in
Here is my take and advice on this…
This is a networking and or resourcing issue on cloud upstream of the portal (you can always see your changes sitting on the cloud portal, just no build). Basically the push to remote dev fails at the build stage because there is an upstream networking issue that I suspect is because there are too many open connections or overall resources are maxed out. This happens if you are deploying via manual or CI/CD streams. I am assuming you are on shared? If so, you can wait a bit, then “touch” either a class file or view and try another push. If you hit it when the server is not busy, you could get lucky.
Advice if you are on shared is to ask support to move you to a less busy app plan and that will solve it until that server fills up with other tenants.
We decided to go to dedicated 1 to make sure we had all the resources available to it and were shocked to find that the error is occurring there too - confirming that the provisioning is too light to handle a dev, prod and SiteExtensions instance on one app plan. We are currently testing a dedicated 2 option and so far no issues further confirming my hunch.
In summary, my guess is that your U16 instance is consuming more resources than the 14 version. If you are on shared then ask to move to a fresh app plan or upgrade to dedicated 2 if you have budget. Most people don’t have budget to go to that level so I am sincerely hoping that the cloud devs find a back-door workaround for this. My understanding is that the service tiers in Azure prevent the ideal configuration required to run cloud as it currently stands…
In my experience this happens almost every time I push up to Cloud, and it’s just a timeout issue when running the pipelines in the cloud (as they maintain the connection during the build).
My go-to is to just check that it’s actually running the pipeline in cloud and just ignore the git error in my client. By the time it’s failed it’s already updated the refs so the local repo isn’t in a broken state.
I had success with setting a few of the commands mentioned on Stack Overflow, but since the build on my projects takes upwards of 10 minutes I decided to just live with the error and keep tabs via the Cloud portal.