Repository navigation
Bump actions/setup-go from 5.6.0 to 7.0.0 - #13
Closed
dependabot[bot] wants to merge 132 commits into
Closed
dependabot[bot] wants to merge 132 commits into
dependabot[bot] wants to merge 132 commits into
Conversation
Added a brief description and a link to pingping.
Added a link to the pingping GitHub repository.
Added a link to the GitHub repository for fogping.
Targets are rows in the existing targets table (new columns: type, host, port, pace, interval_sec, active). The DB is the single source of truth; there is no second copy that can drift or conflict. - --edit enables POST/PUT/DELETE /api/targets and the ✎ targets panel. Default is read-only (writes answer 403 with the reason). --edit refuses to start without a login or --localhost: an open writable UI would let anyone make this host ping/connect to arbitrary addresses. - Writes require Content-Type: application/json and a same-origin Origin (if sent), which closes the cross-site form/CSRF path. - targets/ping.list and tcp.list become an import inbox: dropped files are upserted by name and archived as *.imported; a file with any bad line is rejected whole and parked as *.rejected. This also migrates existing installs on first start; history carries over because ids are looked up by name. - Delete = deactivate: probing stops, history stays until retention; re-adding the name restores it. Rename keeps id, so it keeps history (the list files could not do this). Renaming onto a deleted target's name is refused. - Runner owns the probe loops keyed by id behind one mutex; replaces the mtime-polling reload and the cfg.Targets data race with /api/targets. - Pills no longer use innerHTML: names are web-writable now, so that was a stored XSS.
Rollup
- Ran only at 00:05, so every window over 24h was empty until the first
midnight and always missed the last 0-24h. Now: catch-up over everything raw
holds on start, then the last two hours every 5 minutes.
- Rollup(since) rewrote the bucket containing an unaligned `since` from partial
data just before retention deleted the rest: permanent truncation (a JST host
kept 535 of 1440 minutes per daily bucket). `since` is now aligned down to a
bucket boundary and never reaches back past the raw horizon.
- One streaming pass holding a single bucket's samples, written in one
transaction: no 2-day map of every sample (~140MB for 50 fast targets), and
readers no longer see buckets with zeroed percentiles mid-rollup.
- Loss is computed from packet sums. Hourly rows used to travel as Rounds with
S=n rounds, which rounded any loss under half a round per bucket to 0%.
- All-lost buckets store NULL percentiles instead of 0 ms.
Daily tier removed: kept exactly as long as hourly, so it saved no disk, and it
was the tier the truncation hit hardest. Dropped on startup; >24h windows read
hourly buckets (300d = 7200 rows).
API/UI: /api/series returns {tier, rounds|buckets}. The chart used to sort a
bucket's [min,p50,p90,p99,max] and take the middle, drawing P90 as the median
line. Hourly views now draw p50, show window P50/P90/P99 as the round-weighted
median of the hourly values marked with ≈, and adapt tooltip and legend.
Also: shutdown flushes via store.Close() before exit (up to ~2s of rounds were
lost); TCP probes resolve once per round and dial the IP (DNS time was being
measured as RTT); JSONL-era leftover comments removed.
Tests drive the real scheduler through three days on a UTC+9 clock with nightly
retention; mutation-checked against the old unaligned `since`.
v1.1.0: targets in SQLite + web CRUD (--edit); rollup correctness; tier-aware chart
README - New top section: quick start, then a full guide — requirements, install with checksum, ICMP permission options, first start (with the real log), argument reference, managing targets (web UI fields, pace table, targets/ inbox semantics), systemd with a dedicated user and the stop/--edit/start routine, reverse proxy (nginx must forward Host for --edit), data/retention/backup, reading the chart, upgrading (incl. from 1.0.x), troubleshooting, building. - The 17 language blurbs no longer promise a plain-text config file. - JSONL comparison with pingping removed; sibling kept as a one-line link. Code - ProbeCfg lost its JSON tags (never serialized; config-file leftover). - Schema comment no longer compares the sample blob to JSON. - user=/passwd= and --flags are accepted in any order (Go's flag package stopped at the first user=..., so a trailing --edit was rejected). - ICMP permission error is in English and points to the README.
README: detailed startup guide; drop JSON leftovers
Targets live in SQLite and change in exactly one place: the web UI under --edit. The targets/ inbox from 1.1.0 was a second write path that worked in read-only mode, bypassing the --edit gate; it and the list parser are gone. - Removed: ingest.go, parseListFile/parseListLine, Store.ImportTargets, Config.TargetsDir, targets/*.example. fogping no longer creates targets/. - The demo target is seeded through CreateTarget. - normalizeTarget, now the only input gate, gets a table test and a fuzz target in place of the list-parser ones. - 1.0.x databases still migrate: legacy rows come out inactive and re-adding the same name restores the history (tested). README documents both upgrade routes (via 1.1.0, or re-add by name). - README: inbox section, echo examples and the echo claim in all 17 language blurbs removed; links and wording updated for 1.1.1.
targets: database only — remove ping.list/tcp.list and the import inbox
…ag table --help (also -h, help) now prints to stdout: what fogping does, where data goes, runnable examples for every common case (first try, login, several users, --edit workflow, --days, background run/stop, --localhost, upgrade), the ICMP permission fixes, and a short flag list. Every command in it was executed as printed. Previously it was the flag package's dump on stderr, which did not mention user=/passwd= at all. A bad flag or bad user=/passwd= now prints one line and "Run './fogping --help' for examples" instead of the whole usage; the --edit refusal shows a working example. README: the argument table is replaced by a "Common commands" section that mirrors --help; the target form table is replaced by example targets.
--help with copy-ready examples; README uses commands instead of a flag table
Added images to enhance the README documentation.
Removed redundant lines and clarified description.
Added description about the tool being a single binary with embedded SQLite and built-in web UI.
Updated README to clarify usage instructions.
…ted routes
docs/API.md writes the rules down; api_test.go enforces them, so a route that
breaks one fails CI.
- Every /api response carries API-Version: 1. Changes within v1 are additive.
- Every error is {"error": "<for a person>", "code": "<for a program>"}, as
JSON, including 401s that used to be text/plain and the router's own 404 and
405 answers.
- Every route names its method. Pages are bound to GET too: the catch-all "/"
used to swallow wrong methods as 404; now they get 405 with Allow.
- POST /api/targets and PUT refuse unknown fields, naming the bad one.
- JSON keys are snake_case (the login form now sends user/pass; the old
User/Pass still works). The compact /api/series format is the documented
exception.
- The endpoint table in docs/API.md is checked against the registered routes
both ways.
The web UI and the Zabbix template keep working: error bodies still carry
"error" as a string, and no path changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
icmp and tcp were spelled out in seven places: validation, the probe loop's switch, the address shown for a target, the targets file in both directions, the demo target, and (still, until the UI step) the page. A new probe type would have had to find all of them. prober.go is now the only place that knows which probers exist. Validation, targetAddr, the probe loop and the targets file ask it, so a new prober is one registerProber call. - The probe loop gets a context that is cancelled when the target stops, so a round in flight can be abandoned (an external module will be killed through it). - A target whose prober is missing writes nothing. The old switch had no default and would have stored a zero round at time 0. No behaviour changes: an old and a new binary give byte-identical results for creating targets (errors included), export, import dry-run and the probing log. Tests: no code outside prober.go spells out a type; a fake prober is picked up by validation, the targets file and the probe loop with nothing else changed; stop cancels a round in flight; a missing prober writes nothing. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
PRAGMA user_version records the schema; migrations run in order, each in one transaction with its version bump. Version 1 is the baseline every 1.x database reaches. A database newer than the build is refused. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
A round that could not be measured (DNS failure, no ICMP socket) used to be stored as total loss, which painted a broken setup as a dead target and fed burst detection. It now carries the reason (err column, schema v2), has S=0, and is excluded from stats, burst baselines, hourly rollups and the down flag. /api/targets and /api/stats report it as probe_errors; the UI shows an amber LED with the reason. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
DNS lookup, TCP dial, the pause between packets and ICMP waits all follow the target's context; a round cut short is not stored. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
GET /api/probe-types lists the registered probers; the type filter, the form and the port field are built from it, so a new prober needs no UI change. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
A prober lists option specs when it registers (string, int, bool, choice; defaults, ranges, required). fogping validates them, stores them with the target (schema v3), exports and imports them, serves the specs at /api/probe-types and builds the editor form from them. Probers read the validated values with OptString/OptInt/OptBool. docs/OPTIONS.md is the contract for external authors. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
…ce vectors The language-neutral standard for declaring a prober and validating its settings: JSON Schema, a normative document, stable error codes, and shared test vectors that every implementation must pass. fogping's own validation is checked against them, and every built-in prober is a valid manifest. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
parseRoundResult treats a prober's round answer as untrusted input: shape, types, counts, sample count and range, one-of measurement or error, size cap. Shared vectors in docs/conformance/round.json; the invocation rules (stdin JSON, no argv or env, hard deadline) are written down as a draft. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
Off by default. With --probes, each ./probes/<type>/ holding manifest.json and an executable 'probe' registers a probe type through the same registry as the built-ins. One process per round: JSON on stdin, JSON on stdout judged as untrusted input, nothing in argv, a fixed PATH only, a hard deadline that kills the process group, an output cap and at most 8 at once. A broken plugin is skipped with a log line. The prober registry is now guarded by a mutex. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
plugins/ holds production-ready probers written to be read: http (Go), dns (Python), smtp (Ruby), ssh (Bash), traceroute (Rust). Each follows the fogping-probe/1 contract, validates its input against a whitelist, bounds what it reads, and keeps lost probes apart from its own failures. Each has its own tests against a fake server and a tutorial README; plugins_test.go installs all five and runs them through the real fogping loader, and a CI workflow runs every language's tests. install.sh puts them in ./probes. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
http (time to first byte of an HTTP(S) request) is now a built-in prober, registered like icmp and tcp: no --probes, no separate process. Certificates are always verified, the host is whitelisted before it is put in a URL, and each probe opens a fresh connection. The targets file format and schema document the http type and its options. plugins/http becomes plugins/ldap: an LDAP prober (connect, anonymous bind, root DSE, with TLS) written in Go with a fake-server test, as the reference for writing a prober in Go. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
… error, trusted probe directories HTTP: the path is checked for characters that would corrupt the request line or start a fragment, response headers are bounded to 64 KB, and a name that does not resolve is a round that could not be measured, not 100% loss. External probers: a probers directory, a prober directory or a probe file that group or others can write to, or that another account owns, is refused; install.sh sets safe modes whatever the umask; the trust model is written down. The LDAP prober's go.mod follows the repository's Go version and its x/ modules are upgraded past the published advisories. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
…s refused, demo seed failure logged - POST /api/login now refuses a non-JSON body with 415 like every other write. The check runs before the login throttle, so a refusal cannot take the address's one login slot; the login page now sends the header it needs. - A body over the limit is 413 too_large, as docs/API.md promises, not 400. - 999.999.999.999 and other all-digit hosts that are not IP addresses are refused at entry instead of being stored as a host name and drawn as 100% loss. - Failing to seed the demo target is logged as a warning. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
No panic in production code. A prober whose option declarations are invalid, or whose name is taken, is refused with an error (a plugin is skipped with a log line) and nothing is registered; our own built-in declarations are checked by the tests, not at start-up. A test keeps panic( out of non-test code. A mistake in an optional integration no longer stops the monitoring: a missing or too-short Zabbix token leaves the stats API off and says why in the log. In fogping, --edit without a login or --localhost starts read-only instead of exiting. Login and listen-address mistakes still refuse to start: nothing may fall back to an open console. docs/API.md states that the stats token is read-only by design and that there are no write tokens. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
An earlier change made fogping start read-only instead. That softened a deliberate security rule: nothing may fall back to, or be mistaken for, an editable console open to anyone. The rule is its own function again, held by a test. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
An earlier change let the program start with the stats API off. A wrong token is a configuration mistake the operator should hear about at start-up, not in a log line they may never read. The panic removal stays: a broken plugin is skipped, it does not stop the program. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
…mary Nothing is loosened; every refusal stands. What changes is how easy the right configuration is to reach: - A refused start lists every problem at once, not the first one. - --check runs the same preparation plus the checks that need the machine (data directory, database version, port, plugins) and reports without starting anything. The database is opened read-only and never migrated; a plugin that would be skipped is a warning, as it is at start. Exit 0 means it would start. - --gen-token=F writes a 256-bit random token to F, mode 600, never overwriting and never printing it. - Every start logs a security summary: where it listens, login, editing, Zabbix, probes. Exposed settings are marked with !. Plugin inspection is split from registration so --check can report on a plugin without loading it. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
…lain HTTP and short passwords - An existing database is copied (VACUUM INTO, consistent under WAL) to fogping.db.bak-vN before its schema is upgraded, since an upgraded database cannot be opened by an older fogping. If the copy cannot be made, nothing is changed and the message says why. - A prober now starts through fogping itself as a launcher that sets resource limits on its own process and then becomes the prober: CPU time no more than the deadline, 1 GiB address space, 16 MiB files, 256 open files, no core dumps. The limits apply to that one process, never to fogping. All five official probers pass under them. - The security summary flags running as root, plain HTTP on a network address, and passwords shorter than 8 characters (warned, not refused), naming every login issue at once. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
… and the modules current A tag such as @v4 can be moved to different code by whoever controls the action; a commit cannot. Each pin keeps its version as a comment. Dependabot opens one weekly pull request per ecosystem for minor and patch updates. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01E4SFMB893s7ejgrJ7qosV4
Bumps [actions/setup-go](https://github.com/actions/setup-go) from 5.6.0 to 7.0.0. - [Release notes](https://github.com/actions/setup-go/releases) - [Commits](actions/setup-go@40f1582...b7ad1da) --- updated-dependencies: - dependency-name: actions/setup-go dependency-version: 7.0.0 dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com>
Author
|
OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting If you change your mind, just re-open this PR and I'll resolve any conflicts on it. |
dependabot
Bot
deleted the
dependabot/github_actions/actions/setup-go-7.0.0
branch
November 20, 2026 00:10
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bumps actions/setup-go from 5.6.0 to 7.0.0.
Release notes
Sourced from actions/setup-go's releases.
... (truncated)
Commits
b7ad1dachore(deps): bump@actions/cacheto 6.2.0 (#771)0778a10Migrate to ESM and upgrade dependencies (#763)924ae3achore: bump version to 6.5.0 in package.json and package-lock.json (#762)e91cc3bBump@actions/cacheto 5.1.0, log cache write denied (#758)4a2405echore: update@types/nodeand@typescript-eslintdependencies to latest versi...78961f6chore: update@actionsdependencies and refresh license cache (#744)4a36011docs: fix Microsoft build of Go link (#734)8f19afcfeat: add go-download-base-url input for custom Go distributions (#721)27fdb26Bump minimatch from 3.1.2 to 3.1.5 (#727)def8c39Rearrange README.md, add advanced-usage.md (#724)Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting
@dependabot rebase.Dependabot commands and options
You can trigger Dependabot actions by commenting on this PR:
@dependabot rebasewill rebase this PR@dependabot recreatewill recreate this PR, overwriting any edits that have been made to it@dependabot show <dependency name> ignore conditionswill show all of the ignore conditions of the specified dependency@dependabot ignore this major versionwill close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)@dependabot ignore this minor versionwill close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)@dependabot ignore this dependencywill close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)