"No such host" for <app>.azurewebsites.net — new function apps get a unique default hostname
Function apps created since roughly mid-2024 get a unique default hostname (<app>-<hash>.<region>-01.azurewebsites.net); the bare <app>.azurewebsites.net is never registered in DNS for them, so "No such host" is a naming problem, not an outage. Read the real hostname from the portal or the deploy log — never derive it from the app name.
Function Apps založené zhruba od poloviny roku 2024 dostávají jedinečný výchozí hostname (<app>-<hash>.<region>-01.azurewebsites.net); holé <app>.azurewebsites.net pro ně v DNS vůbec neexistuje, takže „No such host“ je problém s názvem, ne výpadek. Skutečný hostname čtěte z portálu nebo z logu nasazení – nikdy ho neodvozujte z názvu aplikace.
Symptom: your deploy pipeline reports success and the function app runs fine in the portal, but https://<app-name>.azurewebsites.net fails DNS resolution entirely (No such host is known / ERR_NAME_NOT_RESOLVED). Easy to misread as "the app does not exist" or "a transient DNS hiccup" — especially when older apps in the same subscription resolve just fine under their bare names.
Cause: Azure assigns unique default hostnames to newly created apps:
https://<app-name>-<random-hash>.<region>-01.azurewebsites.nete.g. myfunc-c4hdbqf2gxe7a9mz.westeurope-01.azurewebsites.net. The bare <app-name>.azurewebsites.net is never registered in DNS for these apps. Apps created before the rollout keep their bare hostname — so one subscription routinely contains both kinds, which is exactly what makes the failure look random.
Fix / rules:
- Never derive the hostname from the app name. Read it from the portal (app Overview → Default domain) or from your deploy log.
- GitHub Actions tip: the
Azure/functions-actionlog lineDeploy logs can be viewed at https://<app>-<hash>.scm.<region>-01.azurewebsites.net/...reveals the hostname even when the app name itself is masked as a secret — drop the.scmsegment to get the runtime host. - Put the full unique hostname into every config that calls the app (client allowlists, webhook URLs, monitoring, other services' env vars).
- When a bare-name call fails DNS but you believe the app exists, that is a naming problem, not an outage — check the unique hostname before restarting or redeploying anything.