feat(doctor): sweep the linked sites, fold check into site:doctor - #1485
feat(doctor): sweep the linked sites, fold check into site:doctor#1485prabowosd wants to merge 3 commits into
Conversation
lerd doctor ends with a [Sites] section: every linked site runs the cheap half of the site doctor and gets one line, the counts plus the first thing wrong, naming lerd site:doctor <domain> for the detail. The framework command checks, composer audit, npm audit and the response-time lookup stay on the per-site command, and one engine's database list is reused across the sweep instead of costing a container exec per site. Validating .lerd.yaml moves into the site doctor as its own check, so every surface runs it and lerd check becomes a deprecated alias. A worker declared in the project's own custom_workers now counts as defined on a plain site too, not only on a custom-container one.
|
Ran the gate on this and it is green, build, vet, gofmt, the full suite and a race run over sitedoctor and cli. On a twenty site machine the sweep costs about two seconds, and the project config check shows up where it should, both in site:doctor and in the alias. One thing to fix before this goes in. In a directory with no .lerd.yaml, lerd check now prints all checks pass and exits 0, where it used to fail with no .lerd.yaml found, run lerd init. Running it one directory too high used to say so, now it says the project is healthy. The alias should still bail when the file is not there. Three smaller ones. The sweep line reads 2 warning, sitedoctor already has a plural helper for that. cachedDatabases holds one global mutex across the lookup so a postgres exec blocks a mysql one, which does not matter at this scale, but the comment says callers only wait per engine, so either the lock goes per service or the comment goes. And validPHPVersion is now the third copy of the same MAJOR.MINOR rule, next to the one in cli and the one in mcp, worth exporting one instead. Checked what folding check in might have dropped, the service installed and db.service validation is covered by required_services, which resolves db.service too, so nothing is lost there. Thanks for taking this one on, it is a good change. |
2 participants
There are three health commands and none of them knows about the others, so anyone who wants to know whether something is wrong has to run all three, and the one people reach for first never looks at their sites.
lerd doctornow ends with a[Sites]section: every linked site runs through the site-level checks and gets one line, the counts plus the first thing that is wrong, with the command to run for the detail. A machine where podman, DNS and the ports are all fine while three sites are failing no longer reports itself healthy.The sweep has to stay quick, so it runs the cheap half of the site doctor and leaves the framework command checks,
composer audit,npm auditand the response-time lookup tosite:doctoron a named site. The one expensive check it keeps is whether the site's database exists, because that is what catches a site 500ing on every request, and it is no longer expensive: one engine's database list is reused for a few seconds rather than costing a container exec per site. On a fourteen site machine the sweep adds about two seconds to a doctor run.lerd checkwas the third entry point and everything it validated is about the project rather than the environment, so.lerd.yamlvalidation moves into the site doctor as its own check. Every surface that runs the doctor validates it now, the dashboard and the MCP tool included, andcheckstays as a deprecated alias forsite:doctorin the current directory. Folding it in also fixed a worker it was wrong about: one declared in the project's owncustom_workerswas only recognised on a custom-container site, so a plain site running one was told it had no definition to match.The reclaim offer the issue asks for is already there, doctor lists it as an optional fix and
--fixrunslerd cleanupafter confirming.Closes #1358