Hernâni Marques 51785031d6

docs(monitoring): load, coins, pay ladder, alt units, SSH fallbacks

2026-07-17 03:04:54 +02:00

9.8 KiB

Raw Blame History

taler-monitoring — test IDs by area

Every check line is numbered per area as AREA-NNN (zero-padded):

Area	Phase script	Meaning
www	`check_urls.sh`	public HTTPS (outside-in)
inside	`check_inside.sh`	containers / processes on koopa
versions	`check_versions.sh`	deb.taler.net available + packages vs trixie
sanity	`check_sanity.sh`	public + server per component
server	`check_server.sh`	SSH host ports / processes
e2e	`check_e2e.sh`	withdraw + pay cycle

Format in output:

[OK]    www-001  exchange /config https://exchange…/config
[ERROR] e2e-012  bank-auth: admin token failed
[BLOCKER] e2e-015  prereq: merchant HTTP 502

IDs are assigned in run order within the area (set_area resets the counter). Optional soft checks still consume a number when they WARN.

www — public URLs (`./taler-monitoring.sh urls`)

ID	Check
www-…	exchange `/config`, currency, alt_unit_names
www-…	exchange `/keys` (+ alt_unit_names soft)
www-…	exchange `/intro/`, `/` (302→intro)
www-…	exchange `/terms` body (not empty / not API error)
www-…	exchange `/privacy` body
www-…	exchange `/terms/` (200 or redirect)
www-…	bank `/config`, currency, alt_unit_names
www-…	bank integration / webui / intro / `/`
www-…	bank `/terms` body
www-…	bank `/privacy` (or `/intro/privacy.html` fallback)
www-…	merchant `/config`, currency, currencies alt_unit_names
www-…	each merchant `exchanges[]` `/config` alt_unit_names
www-…	merchant `/intro/`, `/webui/`, `/`
www-…	merchant `/terms` body (dual-currency notice)
www-…	merchant `/privacy` body (must not be `not configured`)
www-…	merchant `/terms/` redirect
www-…	landing exposed links (bank / merchant / exchange): parse each `/intro/` HTML, probe every own-stack `https://` + root-relative `href`/`src`/`content`, soft-check external stores/docs
www-…	landing static: `qrcode.min.js`, `og-goa-shop.png`, `qr-logo.png`, shop-pay.js/css
www-…	cross-links between bank ↔ merchant ↔ exchange intros (local stack)
www-…	bank `/intro/demo-withdraw.json` → `taler://withdraw/HOST/taler-integration/…` (no default `:443`/`:80`; non-default port OK) + integration op HTTP 200
www-…	bank `/intro/auto-account.json` (earlier) → same withdraw shape, no payto_uri, login at `/webui/`
www-…	performance (outside-in): public HTTPS RTT for bank `/config`, `/taler-integration/config`, `/webui/`, `/intro/`, `stats.json`; exchange `/config`, `/keys`, `/intro/`; merchant `/config`, `/webui/`, `/intro/` — report ms; WARN ≥ `PERF_WARN_MS` (default 8000); ERROR ≥ `PERF_FAIL_MS` (default 20000)

Legal docs rule: HTTP 200, non-empty body, not plain not configured, not merchant API JSON code:21. On local stack, optional content needle (terms/privacy/FADP/GOA…).

Performance rule: Measured from the monitoring runner (public URLs via Caddy), not container loopback. HTTP must match expect (usually 200); latency (ms) is on every perf OK/WARN/ERROR line, plus a perf summary (n / min / p50 / avg / max). Slow ≥ PERF_WARN_MS (default 8000) → WARN; ≥ PERF_FAIL_MS (default 20000) → ERROR.

Landing links rule: Own-stack (bank/exchange/taler.* + page host) must be HTTP 200 (or redirect→200). External (App Store, Play, F-Droid, wallet.taler.net, docs/git.taler.net, …) soft WARN if down. Auto-account wallet link must be taler://withdraw/HOST/taler-integration/… (default ports stripped for mobile wallets), never payto.

alt_unit_names rule: wallet codec requires a non-empty map including scale key "0". For multi-currency merchant, also follow every entry in exchanges[] and check that exchange’s public /config.

(IDs after a failed early check may shift if later soft checks are skipped when body missing — numbering follows executed checks.)

inside — koopa SSH (`./taler-monitoring.sh inside`)

ID	Check (typical order)
inside-001	ssh koopa (or koopa-external fallback)
inside-002+	per-component emit: container, ports, libeufin/httpd, postgres, local `/config`/`/keys`, wirewatch, DNS pin, caddy
inside-…	load / memory: host loadavg + RAM; bank/exchange/merchant podman CPU/mem/block + process RSS by role + DB sizes

Remote lines E|comp|LEVEL|key|detail each become one numbered result.

SSH: KOOPA_SSH (default koopa), then KOOPA_SSH_FALLBACKS (default koopa-external) when LAN is unreachable.

sanity — bank · exchange · merchant (`./taler-monitoring.sh sanity`)

ID	Section
sanity-001…	bank public + server
sanity-…	exchange public + server
sanity-…	merchant public + server

Sequential through the whole script (one set_area sanity).

versions — packages vs deb.taler.net (`./taler-monitoring.sh versions`)

Outside (runner / public network — no SSH)

ID (order)	Check
versions-…	DNS `deb.taler.net`
versions-…	HTTPS portal + apt base URL
versions-…	suite `InRelease` / `Release`
versions-…	suite `Packages` + `Packages.gz`
versions-…	sample pool `.deb` fetchable (Range 200/206)
versions-…	suite offers `taler-exchange`, `taler-merchant`, `libeufin-bank`
versions-…	optional `trixie-testing` Packages
versions-…	TLS verify (soft)

Inside (SSH koopa containers)

ID	Check
versions-…	ssh koopa
versions-…	each container → `InRelease` (pasta can reach apt repo)
versions-…	each container lists `deb.taler.net` in apt sources
versions-…	each installed `taler` / `libeufin` / `libtaler` / `libdonau` vs suite version
versions-…	core packages installed (`taler-exchange`, `libeufin-bank`, `taler-merchant`)

Outside always runs. Inside skipped with SKIP_SSH=1 (still reports outside results).

Compare rules:

match suite → OK
ahead of suite (often testing/dev) → INFO
behind suite → ERROR for core packages, WARN otherwise (TALER_PKG_BEHIND=error forces ERROR)

Default suite: trixie (TALER_APT_SUITE, TALER_APT_BASE=https://deb.taler.net/apt/debian).

Without SSH (SKIP_SSH=1 or remote domain): still runs outside-in repo checks; skips container install compare.

server — SSH ports (`./taler-monitoring.sh server`)

ID	Check
server-001	ssh
server-002+	containers, local pasta ports, processes, caddy

e2e — payment path (`./taler-monitoring.sh e2e`)

ID	Step (approx.)
e2e-001	budget info
e2e-002	wallet-cli present
e2e-003	mode / currency info
e2e-004…	secrets, reachability gates
e2e-…	account, credit, withdraw, confirm, coins, order, pay ladder
e2e-…	load snapshots at e2e-start, after-withdraw, after-pay, after-shop, e2e-end (host loadavg/RAM + container RSS/CPU/DB)
e2e-…	coins before/after each ATM withdraw, settle, pay, shop: count in circulation / spent, amount + denoms using `alt_unit_names` (e.g. `5 Kilo-GOA (GOA:5000)`), Δ vs previous snap
e2e-…	GOA shop products — full catalog list; random pick of 2 (override `E2E_SHOP_PICK_N`)
e2e-…	balances, dig on failure, load delta + overall metrics

Shop product pays use instance goa-shop (default) and catalog E2E_SHOP_PRODUCTS (id|Product name|amount lines). Each e2e run shuffles the catalog and pays only E2E_SHOP_PICK_N products (default 2). Flow matches the landing popup (public POST /templates/{id}, not private orders). Report labels use product name.

Blockers keep the same ID prefix: [BLOCKER] e2e-0NN step: message.

ladder — GOA withdraw ranges (`./taler-monitoring.sh ladder`)

Step	What
ladder-001	`GET /intro/auto-account.json` (personal account, GOA:0)
ladder-…	load snapshot before withdraws (host + bank/exchange/merchant mem/CPU)
ladder-…	Phase A withdraw (23): mint explorer pool — 0 → random → max−1 → max into one cumulative wallet (mids × `LADDER_WITHDRAW_SCALE`, default 1.5)
ladder-…	wallet-cli accept-uri + explorer confirm when `selected`; settle; coins snap
ladder-…	Phase B pay (23): same shape 0 → random → max−1 → max via merchant orders + `handle-uri --yes`
ladder-…	load snapshot after + delta + overall timing (withdraw + pay)
report	withdraw TSV + pay TSV + JSON

Default amounts (build_ladder_pair, strictly increasing):

withdraw: [0] + random mids + [max−1] + [max]
pay:      [0] + (withdraw_mid / scale) + [max−1] + [max]   # always ≤ matching withdraw mid

TSV range: pin:0 | random | pin:max-1 | pin:max.

Soft: absolute max mint/pay may CEILING_REJECT / skip (WARN). max-1 is a hard pin.
LADDER_PAY=0 skips phase B.

Env	Default	Meaning
`LADDER_STEPS`	`23`	total rungs (0 + mids + max)
`LADDER_MAX_AMOUNT`	`4503599627370496`	absolute last pin
`LADDER_TIMEOUT_S`	`3600`	large rungs need time
`LADDER_MAX_RUNGS`	`99`	cap list length after build

./taler-monitoring.sh ladder
LADDER_STEPS=10 ./taler-monitoring.sh ladder
LADDER_REPORT_DIR=/tmp/my-ladder ./taler-monitoring.sh ladder

Run one area

./taler-monitoring.sh urls      # www only
./taler-monitoring.sh inside    # inside only
./taler-monitoring.sh versions  # deb.taler.net + package drift
./taler-monitoring.sh e2e       # e2e only
./taler-monitoring.sh ladder    # GOA amount ladder + timings
./taler-monitoring.sh -d taler.net urls

9.8 KiB Raw Blame History Unescape Escape

taler-monitoring — test IDs by area

www — public URLs (./taler-monitoring.sh urls)

inside — koopa SSH (./taler-monitoring.sh inside)

sanity — bank · exchange · merchant (./taler-monitoring.sh sanity)

versions — packages vs deb.taler.net (./taler-monitoring.sh versions)