syntaqxA deploy no longer fails because Cloudflare's API stumbled once, and when one does fail it says why. Each of the last few deploys failed at one service's `wrangler deploy` (the runner, then agents and models), and the log showed only the trace of the request that failed, cut off before its reason. Now a `wrangler deploy` that fails for anything but a mistake in what is being deployed (a build that fails, a configuration that does not read) is tried again after 20 seconds and then a minute, which is harmless since deploying the same version again changes nothing, and each failure is reported by Wrangler's own error block, which carries the reason, rather than its last lines.14f96e9History
