Passing the monitor's own port to ssl-checker exposed an adjacent gap found
in review: a plain-HTTP monitor on a non-default port, such as
http://host:8080, now had a TLS handshake attempted against a port that
serves no TLS. That surfaced to the client as a 500 carrying a raw OpenSSL
error ("packet length too long") instead of the previous silent "N/A".
The endpoint never checked the scheme — it was already wrong for http://
monitors, just quietly so, since every such lookup went to port 443. Guard
on the protocol and fail with a 400 before opening a socket.
Adds tests for both plain-HTTP cases (default and custom port), each
asserting the checker is never invoked. Verified by removing the guard:
both fail without it.
fetchMonitorCertificate parsed the monitor URL but passed only the hostname
to ssl-checker, which then defaulted to 443. A monitor on a non-standard
port such as https://host:54321 therefore had the certificate of whatever
answered on 443 inspected instead of its own, surfacing as "N/A" in the
certificate expiry box when nothing was listening there.
URL.port is an empty string for default ports, so it is mapped to 443
explicitly rather than passed through Number(), which would have yielded
NaN. ssl-checker already accepts the port via SSLOptions, so no dependency
or signature change is needed.
Adds a regression suite covering the reported case (non-standard port), the
default-port fallback, an explicit :443, and the NaN guard. Verified by
reverting the fix: the three port assertions fail without it.
Fixes#3646
Requested by a user who found xx.xx% more trustworthy than xx.x% on their
status page. The presentational argument holds, but the stronger reason is
that one decimal was actively misleading.
Intl rounds to the displayed precision, so every value above 99.95% rendered
as "100.0%" — a claim of zero downtime for a period that contained real
outages. On a public status page that is the opposite of reassuring: the
number reads as flawless during a month with a visible incident.
One decimal also collapsed the range operators care about most. 99.9%,
99.95% and 99.99% are three very different commitments (8.7 hours, 4.4 hours
and 52 minutes of annual downtime) and the first rendered as 99.9% while the
other two were indistinguishable at 100.0%.
Everything flows through the shared formatter in FormatUtils, so this covers
both surfaces at once: the monitor list and diagnostics gauges in the web
app, and the uptime figure, chart cells and hover tooltip on status pages.
Infrastructure gauges are untouched — they format separately at zero
decimals, where CPU at "43.00%" would only add noise.
Server-side precision was already sufficient; the aggregation divides
without rounding, so no API change is needed.
Adds tests pinning the regression: 7 of the 10 fail against the previous
one-decimal formatter.
Note: this restores the client Vitest setup. PR #3844 introduced it, but the
config, scripts and logger tests were dropped when that branch merged into
develop, leaving the client with no test runner again.