Description
Is there an existing issue for this?
- I have searched the existing issues
Output of ddev debug test
Expand `ddev debug test` diagnostic information
[COPY-PASTE HERE THE OUTPUT OF `ddev debug test`]
Expected Behavior
The default healthcheck timeout for containers is 120s, but with ddev-webserver since v1.22.0, the actual startup of the php-fpm and nginx processes is only after the mutagen sync is complete. So the 120s should start only after start.sh begins.
Actual Behavior
I think the mutagen sync is inside that 120s, so for large projects (or traditional Windows) this can run over the 120s.
I think this may be why we have such a hard time getting a successful run on traditional Windows (although we have trouble with WSL-docker-desktop as well, so it's also due to Docker Desktop for Windows being a bad thing.)
Steps To Reproduce
Watch the tests.
An example of many, many failures is https://buildkite.com/ddev/ddev-windows-mutagen/builds/4230
Anything else?
I think this can be fixed by changing the healthcheck values (perhaps startup?); and of course we have control of the healthcheck.sh.
A very poor but very cheap approach would be to just increase the timeout on traditional Windows.
Note that this may be affecting projects on macOS as well, where there are super-large codebases, or where large files are accidentally being synced.