Files
Piik/docs/deployment.md
T
AAA 89a03e6470 fix(app): improve capture selection and media recovery (#433)
## Problem and outcome / 问题与结果

This maintenance phase fixes failures around native capture startup, room reconnection and media handoffs. Capture mutations stay cancellable, relayed-content failures remain request-scoped, and failed media connections return to the existing route controller. Windows Auto codec selection measures delivered cadence instead of timer-sensitive per-frame work estimates.

The phase preserves P2P-first routing, one committed graph and one route operation. It adds no signaling replay mechanism, changes no wire/schema version, and includes no interaction prototype or 4K feature work.

## Release notes

### 中文

Piik v1.6.8 是一次稳定性修复,重点改善 Windows 原生采集、断线恢复和分享过程中的控制响应。

#### 修复与改善

- **更准确地选择编码器。** Windows 的自动模式按实际出帧表现评估 H.264,并正确区分启动预热和持续性能不足,减少计时误差导致的不必要 VP8 回退。硬件编码事件改为异步等待。
- **改善断线与线路切换。** 房间信令短暂断开后保留观众成员身份 20 秒;修复线路切换、SFU 订阅恢复及声音重协商卡住的问题。失效连接及时交回现有路由控制器处理,原生 SFU 发送端不再把短暂断连直接判为终止。
- **保持 App 控制响应。** 开始分享和更换来源时仍可取消操作;合法消息突发不再直接关闭控制连接。中继信令的单条内容错误不会连带结束整场原生分享。
- **修正设备列表与配置保存。** 来源或麦克风数量达到上限时保留可用列表;配置替换失败时保留原有设置。
- **改善紧凑布局和辅助功能。** 调整手机横屏下的画面与播控布局、画质预设标签和长邀请链接显示,修正部分提示与辅助功能语义。
- **加强站点访问保护。** 站点登录 Cookie 使用独立随机签名密钥,应用自身阻止被其他页面嵌套。发布说明也会在合并前检查。

#### 升级说明

通信协议和持久化格式保持不变。设置了站点密码时,App 或 Server 重启后需要重新输入站点密码;持久化房间和邀请授权不受这项 Cookie 调整影响。Piik 操作页面不再支持嵌入 iframe。

### English

Piik v1.6.8 is a stability update focused on Windows native capture, connection recovery and responsive sharing controls.

#### Fixes and improvements

- **More accurate codec selection.** Windows Auto mode evaluates H.264 using delivered frames and distinguishes startup warmup from sustained performance limits, reducing unnecessary VP8 fallback caused by timing noise. Hardware encoder events now use asynchronous waiting.
- **Improved reconnection and route handoffs.** Viewer membership survives brief signaling outages for 20 seconds. Fixes cover route activation, SFU subscription recovery and stalled audio renegotiation. Failed connections return to the existing route controller, and the native SFU publisher no longer treats a transient disconnect as immediate termination.
- **Responsive App controls.** Starting a share and replacing its source remain cancellable. Valid message bursts no longer close the control connection, and an invalid relayed payload cannot terminate the entire native sharing session.
- **Device lists and saved settings.** Source and microphone enumeration retain a usable list at their limits. Failed configuration replacement preserves the previous settings.
- **Compact layouts and accessibility.** Improved picture and playback-control layout on landscape phones, quality preset labels and long invitation links, with corrected tooltip and accessibility semantics.
- **Stronger site-access protection.** Login cookies use an independent random signing key, and the application prevents embedding by other pages. Release notes are also checked before merge.

#### Upgrade notes

Wire protocols and persistent formats are unchanged. If a site password is configured, restarting the App or Server requires entering it again; this cookie change does not revoke persisted rooms or invitation grants. Piik operation pages can no longer be embedded in an iframe.

## Verification / 验证

- `npm run check`: passed, 76 test files / 1,171 tests, TypeScript and production Web build.
- `npm run check:go-core`: passed; full Go/native checks also passed during integration of the capture changes. macOS cgo builds remain native-runner checks.
- Windows native Host gates: Auto, H.264 and VP8 passed during the phase; final control changes rechecked with Auto, including live/paused quality, source replacement, failure retirement and subsequent sharing.
- Embedded SFU browser VP8 recovery: passed with production Host/Viewer route owners.
- Embedded SFU native H.264, two rooms: passed, including recovered decoded video/audio, independent room retirement and resource cleanup.
- Documentation, repository hygiene and diff checks: passed. Final follow-up changes only align operations documentation with the PR note check.
- Go race checks could not run locally because no cgo/GCC toolchain is available; the existing Linux CI race job remains required. AMD/Intel hardware, broader device combinations, public-NAT field reports and long endurance are not established by these local checks.

## Impact / 影响

Compatible patch release after v1.6.7. The cookie and frame-protection effects are disclosed in the public upgrade notes. No automatic retry budget, room-code admission policy, room lifetime or media topology is replaced. Reports without matched diagnostics and unmeasured native performance costs remain in the TODO ledger.
2026-09-29 08:44:17 +08:00

12 KiB

Server Release And Recovery

This file owns routine immutable Server releases. Initial service setup is in self-hosting operations; environment and ports are in configuration reference. Exact production identity is owned by the immutable release descriptor, runtime REVISION, and deployment record. Status owns only the compact current product and operational snapshot; it is not a per-release ledger.

The tracked release-server.sh updates an existing Go deployment with embedded media, bare-metal nginx, nftables, and application port 8787. It requires an existing running Go release. Another proxy, firewall owner, application port, first installation, or initial embedded-media cutover needs its own scoped bootstrap/updater; do not call this wrapper generic. Its firewall check compares the nftables ruleset without traffic counters; it does not depend on a particular table name. Coordinate other firewall writers outside the cutover window. See nft output options.

Release Boundary

A routine Server release changes only the built Piik server binary and the Browser assets it embeds. It does not change infrastructure, service units, proxy/firewall rules, secrets, media listener configuration, or persistent room state.

Any task that touches one of those excluded surfaces must first record its exact pre-change value and a recovery procedure for that surface. Do not complicate the routine Server wrapper with one-off infrastructure branches.

The pre-cutover Server release is guaranteed only through bounded health and postflight. It has no retention contract afterward and is not a maintained backup; Git and immutable artifacts own history.

Build Host

Start from a clean exact revision with Node.js 24, npm 11, and Go 1.26:

npm ci
npm run check
node scripts/package-server-release.mjs <output-directory-outside-repository>

The packager refuses a dirty tree, builds the Browser assets and the linux/amd64 server binary from that exact revision without cgo, records the full revision, emits a runtime archive plus path/size/SHA-256 manifest and release descriptor, and extracts its own artifact to verify it. The archive contains exactly piik-server, LICENSE, THIRD-PARTY-NOTICES.txt, and REVISION. Upload the archive, manifest, and descriptor together to /opt/piik/uploads.

Public Release pages expose the runnable packages and put package checksums in the release body. Retain the original complete build output outside the checkout for later managed deployment or mirror retries: descriptors and manifests are available in CI artifacts for 14 days and are not public Release attachments. Do not rebuild an existing published version to replace missing metadata or bytes.

The same Server descriptor is also the App assembly input. On each target platform, provide that platform's Go toolchain:

PIIK_GO=/path/to/go node scripts/assemble-app.mjs \
  /outside/repository/server-release/piik-<revision>.release.json \
  /outside/repository/piik-app \
  --target windows-amd64 \
  --capture /path/to/platform-capture \
  --tunnel /path/to/cloudflared

Assembly refuses a dirty or different revision and emits one directory with the App executable, the Browser assets taken from that release and embedded in it, the selected sidecars, notices, and matching REVISION. Its required target is one of windows-amd64, linux-amd64, or darwin-arm64; every supplied binary input must match it. It does not create an installer, auto-updater, release tag, or compatibility bundle.

Darwin App assembly runs on macOS with its SDK and cgo enabled: the pinned LiveKit dependency's Darwin CPU statistics call Mach APIs. Windows and Linux App builds keep cgo disabled. A core check on those hosts explicitly skips the Darwin binaries; macOS build and package verification belong to the existing native runner. This does not change the cgo-free linux/amd64 Server artifact.

The platform package also carries its native presentation metadata: Windows embeds the icon in the Go executable, Linux emits a freedesktop desktop entry under share/, and macOS emits a thin .app launcher with an ICNS resource. The latter two are packaging metadata only and do not duplicate the App executable.

CI and local release-candidate builds use the same wrapper on the target's native operating system:

node scripts/package-app-candidate.mjs \
  /outside/repository/server-release \
  windows-amd64 \
  /outside/repository/app-candidate

The wrapper downloads the pinned public-link sidecar, verifies its digest, builds the target App and available capture process, executes every packaged runtime from the extracted ZIP, and emits one .zip plus its SHA-256 file.

App ZIPs use ordinary Store/Deflate compression and relative entry names without . or .. path segments; include hidden files and preserve executable modes on Unix. Windows packaging also checks every entry through Windows Explorer's ZIP reader. A successful extraction by the writer's own library does not establish compatibility with the user's default archive tool.

Retain the descriptor and successful deployment output as release metadata. Do not create a follow-up source commit solely to duplicate their revision, asset, or hashes.

Manual app_checks=true dispatch packages Server and three-platform App candidates after validation. Once automatic publication is explicitly enabled, accepted main merges run that same pipeline and publish the verified artifacts. Candidate artifacts are retained for 14 days. Ordinary branch pushes stay quiet; PRs run only the read-only release-note check. No release-record commit is written back to main. The workflow does not deploy into a running application service.

Do not build or run the full repository check on a constrained production host. That host runs only the packaged binary and needs no Node, npm, or dependency install.

Update Check

Builds carry a product version plus full source revision. Packagers use one release plan or exact Git tag and inject its identity into Server, App and Web; untagged local candidates use development. Schema-2 package descriptors carry the same pair and artifact hashes. The version policy owns ordering and first-public-release readiness.

The default App launcher starts immediately, then checks GitHub Releases in the background, falling back to the Gitee mirror if GitHub is unavailable. The release-source policy owns selection and provenance checks. It shows a link only when the latest stable release is newer, the same version has a known different source SHA, or a development build can choose the official release. These notices are distinct; a different SHA alone is not called newer. The request sends no current revision, credentials, room data, or media data; network errors, private-repository responses, and missing releases are treated as no notice. Checking does not download an archive, replace files, or interrupt a running share; the user can follow the update link to download the App ZIP.

An operator can perform the corresponding read-only Server check:

bash deploy/check-release.sh

The Server also checks once in the background after startup and logs an available release with its download URL. It does not delay startup, periodically poll, notify room participants or install anything.

The script runs the deployed binary's --check-release, using that binary's injected identity, and prints one JSON result with versions, available source SHAs and the release URL. Exit 0 means no newer/different official build, 10 means update-available, different-build or official-release, and 20 means the check is unavailable. It does not accept a separate revision file. For a private repository, inject a short-lived GITHUB_TOKEN through the operator environment; never place it in the repository or command line. The command does not mutate files, services, containers, or persistent state.

Atomic Cutover

The wrapper requires an already running packaged Go release and the service layout described above. For a new installation, use the self-hosting guide. Infrastructure changes need their own scoped recovery procedure under CONTRIBUTING.

Run the tracked server entry with the uploaded descriptor:

sudo env PIIK_PUBLIC_ORIGIN=https://share.example.com \
  bash /path/to/repository/deploy/release-server.sh \
  /opt/piik/uploads/piik-<revision>.release.json

The wrapper:

  1. takes a deployment lock and validates the descriptor, revision, archive, manifest, current Go process, Piik/nginx service state, and public origin;
  2. rejects unexpected archive paths, links, file types, or duplicate inodes;
  3. extracts to a new release directory and verifies every file hash and size;
  4. validates production configuration by running the packaged binary's configuration check as the service user under a bounded runtime;
  5. proves the old and new releases share no regular-file inode;
  6. atomically switches /opt/piik/current, starts the service, and polls bounded local/public health; and
  7. restores the exact prior symlink and service when cutover or health fails.

Never hard-link release files or recursively mutate permissions before the inode audit: metadata changes would violate both releases. Expected service states should be read as data (systemctl show), not used as bare commands under strict-shell error traps.

Persistent State

Stable mode stores only room authority at the configured SQLite path. The systemd service owns its state directory and file permissions. Application restart retains room ownership, invitations, policy and password verifier; participants, signaling, routes, SFU state and media reconnect from fresh process state.

Lightweight mode has no persistent room state and starts empty. A release does not add a migration or compatibility reader for either mode; current exact schema validation fails closed before accepting signaling or media admission.

Postflight

Verify only the surfaces relevant to the release:

  • local and public /healthz return {"status":"ok"};
  • the public HTML references the new immutable main asset;
  • Piik and nginx have expected active/restart state;
  • Piik owns each configured STUN/SFU UDP listener and public reachability matches the configuration;
  • the current symlink and served revision match the descriptor;
  • room creation, Host authentication, one direct Viewer, and configured SFU fallback complete the scoped smoke; and
  • infrastructure, config, secrets, ports, and persistent state not named by the task are unchanged.

Representative network, mobile, audio, 20-Viewer, and endurance evidence belongs to verification status, not routine postflight.

Recovery

Before cutover, the wrapper keeps the exact prior release path. On failure after the switch begins, it stops Piik, restores that path atomically, restarts the service, and waits for health. A failed recovery exits distinctly and requires operator intervention; it must not silently report the new release as active.

Infrastructure or persistent-state recovery is separate and limited to the surfaces changed by that task. Do not restore an entire server, old firewall, or old secret set for an application-only failure.

If a task changes persistent room data, prepare and verify a separate SQLite backup and recover matching data, configuration and application together. Restoring old data can revive revoked credentials; account for later authority changes before reopening ingress.