jarl block pause|resume (and the Blocking page) lets every name through
for a while; jarld resumes on time. The lists stay loaded: the rules zone
just passes everything. Statistics gain the clients with the most blocked
queries and blocked queries on the queries chart. The view choice in the
blocking forms is a dropdown.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
jarld downloads blocklists (Adblock, hosts or plain domains), checks them
daily and reloads named only when a list changed; your own block and
allow rules take precedence, and the most specific rule wins. Lists and
rules apply per view (default: every view that offers recursion). Web UI
Blocking page with name checks, jarl blocklist and jarl block, the
/blocking API, blocked counts in the statistics. Database schema 6.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The zones list shows a padlock after signed zones, linking to their
DNSSEC tab. Client lists in the statistics show the host name Jarl's
records have for each address (PTR, else A/AAAA); no DNS queries.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
make deb now builds two packages: jarl-cli (the CLI, its completion and
the ACME hooks, no scripts) and jarl (the server), which depends on
jarl-cli of the same version.
Pushing a v* tag runs .gitea/workflows/release.yml: it tests, builds the
tarballs, the CLI for Linux, macOS and Windows (make cli-dist) and the
.debs for amd64 and arm64, then scripts/gitea-release.sh creates the
Gitea release and uploads the .debs to the Debian package registry, so
servers install and upgrade with apt. The stats package's inode lookup
is split by OS so the CLI builds for Windows.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Off by default (stats in jarl.yaml). When enabled, Jarl adds a loopback
statistics channel and a response log to named's configuration (merged
into existing logging/statistics-channels statements) and switches
response logging with rndc responselog, because the responselog option
only counts when named starts. jarld polls named's counters every minute
and tails the response log (resuming after restarts and across
rotation). It stores aggregates, never individual queries, in a separate
SQLite database: per-minute, hourly and daily counters (queries, result
codes, record types, views, zones, cache hits, upstream latency, DNSSEC)
and hourly/daily top lists (names, clients, NXDOMAIN and SERVFAIL names,
refused clients), each with its own retention. Clients can be anonymized
or excluded.
- Web UI: Statistics page with KPI tiles, queries and failed answers over
time, a weekday x hour heatmap, top lists and breakdowns; SVG charts
with hover/keyboard readouts, a table view and light/dark palettes
validated for colour-vision deficiency.
- CLI: jarl stats status|summary|top|timeline|heatmap|breakdown.
- API: /api/v1/stats/*, behind a new "stats" permission resource.
- jarld brings named's configuration up to date with jarl.yaml at
startup (engine.SyncConfig), never overwriting hand-edited files.
- resync now refuses to overwrite drifted files without force, as
documented.
- tmpfiles.d creates /var/log/named (allowed by named's AppArmor profile
on Debian 13 and Ubuntu).
- The e2e harness runs named with -f like the Debian unit.
- docs/stats.md, plus API, CLI, configuration, permissions and README
updates.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
adamandClaude Opus 5.5
abrown38andClaude Opus 5.5