Skip to content

Releases

isitmyisp 0.5.2

Choose a tag to compare

@21Enzo17 21Enzo17 released this 18 Nov 00:12
72d76bf

Fixed

  • Zombie mtr processes in Docker: every path re-discovery (every 15 min) left one <defunct>
    process behind, because mtr exits without waiting for its helper and the orphan landed on
    isitmyisp as PID 1. isitmyisp now reaps them, and the image runs under tini (#94).

Changed

  • The installer and the manual docker run cap the container log at 3 files of 10 MB.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.5.2

isitmyisp 0.5.1

Choose a tag to compare

@21Enzo17 21Enzo17 released this 17 Nov 04:18
7076351

Security

  • Open redirect in ?next=: /theme?next=/%09/evil.example (no login needed) sent the browser
    to another site, because browsers drop tabs and newlines from a URL; a newline gave a 500. The theme
    switch and the login now accept only a path on this site (#91).

Added

  • FAQ entries on how the verdict is reached, the container's permissions, the installer and how the
    project is written with AI (#92).

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.5.1

isitmyisp 0.5.0

Choose a tag to compare

@21Enzo17 21Enzo17 released this 17 Nov 03:18
0262d13

Added

  • The TV board shows more at a glance: the doctor's network grade (in place of the 24 h
    availability once the doctor has run), 30 days of internet and ISP availability, the last speed
    test against your plan with where the bottleneck is, and call quality. Each block shows only when
    it has data; ?layout=simple is unchanged (#84).

Changed

  • Settings in four tabs (Monitoring, Your connection, Alerts, Access), still without JS. The
    contracted plan has its own card, and switched-off features fold away until you open them (#85).

Fixed

  • A day with measured windows could show as "before monitoring" in the 30-day strip when the first
    window ended just past midnight (#87).
  • Without the web extra, the web tests used to skip themselves and a broken install still passed;
    they now fail unless ISITMYISP_CORE_ONLY=1 says the core-only run is on purpose (#83).

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.5.0

isitmyisp 0.4.0

Choose a tag to compare

@21Enzo17 21Enzo17 released this 16 Nov 19:43
f029f60

Added

  • Loops and broadcast storms on your LAN get a verdict of their own: duplicate replies, broadcast
    well above normal, or the router's MAC moving. With the router over SNMP, its broadcast and
    multicast counters too. A loop demo scenario (#64).
  • isitmyisp doctor, a check-up of how your network is set up even when nothing is down: CGNAT,
    double NAT, IPv6, DNS speed and hijacking, MTU, router reboots, loops, link speed, NIC errors and
    Wi-Fi. Green, yellow or red with what to do, and an A–F grade next to the list. It also runs once a
    day, has its own dashboard page and goes in the report (#65).
  • Check from this device (/check): measure from a PC or phone in about 15 seconds, crossed with
    what the probe saw in the same minutes. The page loads /static/check.js, allowed by its CSP only
    there, with connect-src https: to reach the internet destinations (#66).
  • Speed inside and outside the ISP: up to three iperf3 servers of your own, marked inside or
    outside your ISP by ASN (or by hand), next to Cloudflare. The dashboard and the report say whether
    the bottleneck is your access or the ISP's exit (#68).

Fixed

  • The speed test confirmation said "about 15 seconds" even with iperf3 servers, each of which adds
    about as much.
  • Docs and README brought in line with the code: the speed test's incident and busy-link guards
    apply only to scheduled tests, the "Measurement" settings card, BBVA in the presets, and every new
    outgoing connection listed in Privacy (#75).
  • The dashboard grid no longer squashes the link, calls and services cards between 761 and 1100 px
    wide (#77).

Changed

  • The sidebar's bottom: the "System active" box is now one quiet status line, and the account,
    language and version live in a menu (#78).

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.4.0

isitmyisp 0.3.0

Choose a tag to compare

@21Enzo17 21Enzo17 released this 12 Nov 21:27
fa9605c

Added

  • The chart, enlarged (/chart, "Enlarge" on the dashboard): drag across it to zoom into that
    stretch and double-click to zoom out, like in Grafana; a range picker with quick ranges (the last
    15 minutes to the last 30 days) and an exact from/to; earlier, later, zoom in and out. Up to 90
    minutes every column is one 30-second window with its exact loss, and a table lists them. The
    dashboard's "Last 24 hours" button opens the same picker.
  • Each monitored destination (your router, the ISP's first hop, each internet destination, DNS)
    opens its own enlarged chart.
  • The dashboard tooltip shows each column's peak loss next to its average: a few seconds at 40 %
    no longer look like 0.3 %.
  • The public demo has every zoom of its day, and dragging there jumps to the closest one.

Changed

  • "Without outages" shows the last 24 hours as a strip (green, and the colour of whoever was at fault
    in each outage) instead of a line of accumulated healthy hours, which said little and could run
    over a long figure.
  • The UI still works with no JavaScript, but it's no longer script-free: the enlarged chart loads
    /static/chart.js (drag-to-zoom), allowed by its CSP only on that page and never inline.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.3.0

isitmyisp 0.2.3

Choose a tag to compare

@21Enzo17 21Enzo17 released this 12 Nov 19:37
b19dfc2

Fixed

  • With Tailscale's MagicDNS on the machine, the DNS server measured was Tailscale's local forwarder
    (100.100.100.100), not your network's. It's now skipped like systemd-resolved's stub. A resolver on
    the same machine (AdGuard Home, Pi-hole on 127.0.0.1) is measured if it answers. When nothing else
    is found, the router is measured only if it answers DNS: many business routers and firewalls don't,
    and measuring one blamed "your DNS" all the time.

Added

  • Settings → General → DNS servers: choose which DNS servers to measure (up to 3, private
    addresses and 127.0.0.1 allowed, for AdGuard Home or Pi-hole). Empty means automatic, and the
    field shows what was found.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.2.3

isitmyisp 0.2.2

Choose a tag to compare

@21Enzo17 21Enzo17 released this 12 Nov 17:01
efe74af

Fixed

  • isitmyisp status crashed (AttributeError: no attribute 'db'): it read --db without declaring
    it. A test now checks that every command only reads arguments its parser declares.

Changed

  • Removed the old TV board's stylesheet rules (unused since the redesign; nothing looks different).
  • scripts/dockerhub-readme.sh puts the README on the Docker Hub page; docs/brand/social-preview.png
    is the card for links to the repository.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.2.2

isitmyisp 0.2.1

Choose a tag to compare

@21Enzo17 21Enzo17 released this 12 Nov 16:35
bc74ffb

Fixes found before announcing it: the verdict missed intermittent outages, and the demo was cut off
on common laptop screens.

Fixed

  • A connection that drops and comes back (bad, good, bad, good…) never opened an incident, so it got
    no alert and no line in the ISP report: it only showed as separate short blips. Three bad windows
    within five minutes now open an intermittent incident, which says how many times it dropped
    on the dashboard, the TV board, alerts and the report (and the ISP message), and closes after three
    clean minutes. An outage whose ISP first hop looks affected only in some windows is no longer split
    into separate "backbone" and "access network" incidents, nor kept from opening.
  • After a restart, what's normal for each destination's latency was learned again from scratch (and a
    slow spell going on at that moment could be learned as normal). It's now loaded from the last day's
    healthy windows.
  • On a 1366 or 1440 px laptop the dashboard's 24-hour card was too narrow and its data table spilled
    over the next card. The chart now takes the whole row below 1800 px, cards stack below 1100 px, and
    the table scrolls inside its card.
  • The report shown in the web app had no way back: it now has "Back to the dashboard" and, outside the
    demo, "Download PDF" (neither is printed nor in the files that get attached or saved).

Added

  • The public demo has TV mode (the board and the simple view) and an intermittent outage in its day.
  • Docs: TV mode, devices and opening hours with screenshots; how it differs from Uptime Kuma,
    SmokePing and PingPlotter; backing up and moving to another machine; when the monitor doesn't
    start or doesn't see the network; what it doesn't do (IPv4 verdict, small latency rises, Linux only).
  • A README in Spanish, and a shorter README in English.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.2.1

isitmyisp 0.2.0

Choose a tag to compare

@21Enzo17 21Enzo17 released this 12 Nov 05:26
35f8e87

Ready to sell: what a small business sees in the first meeting, and the image published for amd64 and
arm64 so nobody has to build it.

Added

  • Opening hours (#34): per weekday with up to two shifts. The dashboard, the TV board and the report
    count the downtime that happened while the business was open, each incident in the report says
    whether it was in or out of hours, and alerts can be held while closed with one summary at
    opening time. A step in the first-run wizard; the public demo has a shop's hours.
  • Monthly report by email (#36): on a chosen day and hour, last month's evidence report goes out with a
    plain-language summary in the message and the report attached (PDF in the Docker image, HTML
    otherwise). Uses the alerts' SMTP server, sent once per month across restarts, retried on failure,
    with a "send now" preview.
  • Devices on your network (#10): card terminals, printers, camera recorders, phone systems, access
    points, checked by ping or a TCP port every 30 s. When one stops answering while the router still
    does, the dashboard, TV board, alerts and report say it's the device, not the internet. Private
    addresses are accepted only for these. The demo has three, and its card terminal drops once.
  • TV mode at the short address /tv: the TV shows a code, an admin types it in Settings, and the screen
    stays paired with its own device cookie (codes expire in 10 minutes and work once; screens can be
    unpaired one by one). The direct secret link remains as an alternative.
  • A light/dark button that works without JavaScript, and a documentation site at /docs (es/en).

Changed

  • The TV board follows the new design: the verdict and three figures on top, the last 24 hours beside
    the latest events, then the destinations as tiles and your services, devices and tunnels. It fits
    any screen from 720p to 1440p without scrolling, shows only events that fit whole, and no longer
    relies on CSS that older TV browsers lack.
  • A new design for the dashboard and every page: a sidebar, the verdict beside three figures with
    sparklines and change against the day before, and cards of equal height for the chart, destinations
    and events.
  • The evidence report reads in two parts: a summary for the owner first (worked fine, outages, problems
    while open, speed against the plan, comparison with the period before, the outages that hit most,
    devices and services in one line each, and what to do), then the technical annex for the ISP, which
    opens with the ready message. It follows the new design, in HTML and PDF.

Fixed

  • Old closed incidents and their mtr captures are now deleted after about 13 months, as documented.
  • The docs builder used an f-string form that needs Python 3.12; it now runs on 3.10 and 3.11 too.
  • Stylesheet URLs carry a fingerprint, so a new design is never stuck in a browser cache.

Install: curl -fsSL https://isitmyisp.enzo-meneghini.com/install.sh | sh
Image: docker.io/21enzo03/isitmyisp:0.2.0