v0.36.11

CHANGELOG.md

@@ -4,9 +4,11 @@ Release history focused on **features, settings, default behavior, compatibility
 
 ## 0.36.11 — more Backends metrics
 
-- **Backends overview — new columns:** seven more metrics, each sortable and switchable under **Columns**, fill the empty space of wide screens. **HTTP codes** (on by default) shows the 2xx share, any 4xx/5xx shares and a stacked 2xx/3xx/4xx/5xx bar. **Cache** (on by default) shows a HIT-ratio ring and how many bytes were served from cache versus fetched from the backend (ORIGIN); the tooltip lists HIT/MISS/BYPASS counts. **Req/s** shows the current request rate of the backend. **p95 / p99** shows latency percentiles, read from the latency histogram (shown as the bucket's upper bound; above 5 s is shown as ">5 s"). **Probe** shows the last active health-check latency and how long ago it ran. **Error causes** shows up to three most frequent failure counters (5xx, timeout, connect, TLS, DNS, I/O, queue, limit, retry). **Protocol** shows the last observed upstream protocol (H1/H2) and the HTTP/2 share. HTTP codes, Cache and p95 / p99 are counted per vhost, so all backends of one vhost show the same value. The vhosts panel backend table offers the same columns as text.
+- **Backends overview — new columns:** seven more metrics, each sortable and switchable under **Columns**, fill the empty space of wide screens. **HTTP codes** (on by default) shows the 2xx share, any 4xx/5xx shares and a stacked 2xx/3xx/4xx/5xx bar. **Cache** (on by default) shows a HIT-ratio ring, the response bytes served from the vhost's cache (CACHE) and the response bytes this backend delivered (ORIGIN); the tooltip lists HIT/MISS/BYPASS counts. **Req/s** shows the current request rate of the backend. **p95 / p99** shows latency percentiles, read from the latency histogram (shown as the bucket's upper bound; above 5 s is shown as ">5 s"). **Probe** shows the last active health-check latency and how long ago it ran. **Error causes** shows up to three most frequent failure counters (5xx, timeout, connect, TLS, DNS, I/O, queue, limit, retry). **Protocol** shows the last observed upstream protocol (H1/H2) and the HTTP/2 share. HTTP codes, Cache and p95 / p99 are counted per vhost, so all backends of one vhost show the same value. The vhosts panel backend table offers the same columns as text.
 - **Status API:** each `vhost_metrics` entry in `GET /api/status` (full and `view=live`) adds `status_classes` (1xx–5xx counters), `p95_latency_ms` and `p99_latency_ms` (`null` without traffic). A new top-level `vhost_cache_totals` object (vhost ID → `hits`, `misses`, `bypasses`, `saved_body_bytes`) sums the cache domain statistics per vhost.
 - **Preferences API:** `backend_table_columns` and `backend_table_sort` accept `rate`, `statuses`, `p95`, `cache`, `probe`, `causes` and `protocol`. The default column set adds `statuses` and `cache`; the other new columns are opt-in.
+- **Info popovers:** the "i" explanations (for example next to Response time history) were drawn as one long line running past the card when they sat in a single-line chart header; they now wrap inside the popover.
+- **Documentation:** refreshed screenshots (dark theme, 90% scale, interface icons, collapsed menu, colour-blind safe charts) and a new Backends card screenshot in the screenshot tour.
 - **Upgrade notes:** no configuration or database migration is required. An account that has already saved its own column choice keeps it, so enable the new columns under **Columns** in the Backends card.
 
 ## 0.36.10 — Fit to screen, collapsible menu and smoother startup

docs/SCREENSHOTS.md

@@ -2,6 +2,8 @@
 
 A tour of the RustProxy administration panel. The panel runs on the separate administration listener (by default `http://127.0.0.1:9090`) and is available in English and Polish, with light and dark themes.
 
+The screenshots show demo data with the dark theme, 90% interface scale, SVG interface icons, the collapsed left menu and the colour-blind safe chart palette.
+
 ## Sign-in
 
 Local administrator sign-in. The first account is created with a one-time setup token; there is no default password.
@@ -14,6 +16,10 @@ Live requests per second, bandwidth, status codes, cache hit ratio, protocol mix
 
 ![Overview dashboard](images/overview.png)
 
+The Backends card lists every backend with request rate, HTTP status classes, latency trend, p95/p99, health probe, cache HIT ratio with cache/origin bytes, traffic and the last event. Columns can be chosen under **Columns**.
+
+![Backends card](images/overview-backends.png)
+
 ## Vhosts
 
 Every published domain with its listeners, backends, health, request rate and security options. The editor groups frontend, backends, balancing, security, cache and custom locations.

src/admin/web/css/10-base.css

@@ -1324,6 +1324,9 @@ html[data-interface-icons="true"] .network-interface-hidden .quiet>svg.ui-icon {
   border-radius: var(--rp-radius-surface, 6px);
   background: var(--panel2);
   box-shadow: 0 12px 30px var(--shadow-mid);
+  white-space: normal;
+  text-align: left;
+  overflow-wrap: break-word;
 }
 
 .metric-info-popover .fine {

src/admin/web/vhosts/render.js

@@ -297,7 +297,7 @@ function backendExtraMetrics(pool, member, status, snapshot, now, seen) {
   const classes = Array.isArray(vhost.status_classes) ? vhost.status_classes.map(value => Math.max(0, Number(value) || 0)) : [0, 0, 0, 0, 0];
   const classTotal = classes.reduce((sum, value) => sum + value, 0);
   const hits = Number(cacheTotals?.hits) || 0, misses = Number(cacheTotals?.misses) || 0, bypasses = Number(cacheTotals?.bypasses) || 0, lookups = hits + misses + bypasses;
-  const cacheBytes = Math.max(0, Number(cacheTotals?.saved_body_bytes) || 0), outBytes = Math.max(cacheBytes, Number(vhost.bytes_out) || 0), originBytes = Math.max(0, outBytes - cacheBytes);
+  const cacheBytes = Math.max(0, Number(cacheTotals?.saved_body_bytes) || 0), originBytes = Math.max(0, Number(member.bytes_out) || 0), outBytes = cacheBytes + originBytes;
   const p95 = Number(vhost.p95_latency_ms) || 0, p99 = Number(vhost.p99_latency_ms) || 0;
   const probeOff = member.health_mode === "off", probeLatency = Number(member.latency_ms) || 0, probeAt = Number(member.last_probe_ms) || 0;
   const causes = BO_CAUSES.map(([field, label]) => ({ label, value: Math.max(0, Number(member[field]) || 0) })).filter(cause => cause.value > 0).sort((a, b) => b.value - a.value);