Repository navigation
Releases
v0.8.0
- Web and desktop: a default logo ships with the app (#90). A deployment without an uploaded logo showed the project name's first letter in the sidebar, on the sign-in page and in the theme-wizard previews; it now shows a default mark tinted with the brand color. The sidebar brand row defaults to logo plus project name when
brandShowis unset; an explicitbrandShowis unchanged. The sign-in page plays the mark's bloom animation once, over five seconds, and not at all underprefers-reduced-motion. Upgrade:WebStack. - Desktop (Windows): a signed install updates itself the way macOS does (#86). Until now Windows only announced a new version and opened the download page. Now, when the installed app carries a publisher name in its
app-update.yml(every signed release, see #37), the installer downloads in the background, electron-updater verifies its Authenticode signature against that name and refuses a mismatch (ERR_UPDATER_INVALID_SIGNATURE; nothing is announced for it, and Check for Updates… shows the error), and the tray row and notification offer Restart to install: that runs the Quit teardown, then the one-click installer silently (per-user, no elevation prompt) and relaunches the app. A tray Quit installs it silently too. Installs of 0.7.0 and earlier carry no publisher name, so they keep announcing only and are replaced by hand once with the next signed installer; from then on they update themselves, which the release after that is the first to show. macOS is unchanged. Nothing in the package changed. - Desktop and web: the app no longer renders frames nobody can see (#87). The desktop app keeps Chromium's background throttling off so credential renewals stay on time in the tray, which also kept every backend view rendering at 60 frames per second while the window sat in the tray, minimized or showing another tab; the GPU process held about 20% CPU and drained the battery. The desktop app now pauses every CSS animation in a view nobody can see and tells the page, which stops the network map's canvas loop; the active tab of a shown window resumes. The background glow behind the shell, the login page and the theme wizard no longer drifts: a layer moving under the panels' backdrop blur made the GPU recompute the blur every frame even in a visible window, for a barely visible effect. Timers are unchanged. Upgrade:
WebStackfor the glow; the desktop part ships with the next desktop release. - Desktop: the update notification no longer goes off once and disappears when a background download happens to finish (#88). It waits until someone is at the machine (screen unlocked, input in the last minute) and repeats once a day until the update is installed; a newer version notifies at once. Clicking it opens the Restart now / Later box (macOS, and a signed Windows install, #86); on an unsigned Windows install it opens the download page. The tray row is unchanged, and Check for Updates… still answers in its own dialog.
- Desktop: the Windows installer is signed by FNTIO GmbH through Azure Artifact Signing (#37), so Windows names FNTIO GmbH as its publisher, and machines whose policy blocks unsigned apps can install it. SmartScreen may still show "Windows protected your PC" until the new signing identity has built download reputation; More info, then Run anyway gets past it as before. Installs of this and later releases also update themselves (#86). Nothing in the package changed.
Full changelog: CHANGELOG.md · npm: @fntio/infra-client
v0.7.0
Upgrade: CredsStack, WebStack; NetworkStack and every ConnectorStack if you run the private network.
-
Private network: a failed relay map refresh no longer leaves a site without relays (#83). Headscale v0.26.1 refreshes the relay (DERP) map from
controlplane.tailscale.comevery 24 hours and, when one download fails, sends every node an empty map until the next refresh, so users could not connect for up to a day. Each control-server start now saves the live map to the site's EFS folder (a failed download keeps the previous copy and never stops the task), and Headscale merges that copy with the live map region by region (the live map wins per region), so a failed refresh still leaves every region of the copy. A region Tailscale removes stays in the map until the next control-server start. Connectors get an ECS health check: one that has reported "could not connect to the ... relay server" for about five minutes is replaced automatically, which is what a stuck connector needed after the map came back. With an alert topic, the newDerpMapLoadFailedalarm fires when a control server logsCould not load DERP map. Upgrade:NetworkStack(manual,--exclusively; restarts each control server once) and everyConnectorStack(restarts each connector once). -
Sign-in: turning on an authenticator app now makes this device remember password as its sign-in method, so the login page stops jumping to a passkey or email code that Cognito no longer offers that user. The Passkeys dialog and the "Only your password is offered" notice now say an authenticator limits sign-in to password + code. Upgrade:
WebStack. -
Users: the People list is sorted by email A to Z, and a Sort dropdown in the toolbar switches it to email Z to A or to last seen (most recent first, or longest ago first; someone who never signed in counts as the oldest) (#80). The sort runs in the browser after search and filters; a reload starts at A to Z again. Upgrade:
WebStack. -
roleSessionName: "userId" | "email"(#65) chooses how a vended role's STS session is named, which is what the target account's console and CloudTrail show as the federated user (<Role>/<session>).userId(the default, and every deployment's behaviour so far) keepscreds-<Cognito sub>;emailnames the session after the caller — a person's email, a personal access token's owner's email, a service user's registry name — so a team in the target account can see who made a change without asking an Infra Client admin to look the sub up. Characters STS does not allow become_; an empty or over-long name falls back tocreds-<sub>. The admin Roles page's session-length Test follows the same policy (test-<sub>or the admin's email). Role trust is unchanged: the signed token still pins the caller's sub. Sessions vended before the switch keep the UUID in CloudTrail. Upgrade:CredsStack. -
The
StaleConnectorsalarm no longer fires for a connector node that a trusted online connector in the same account and VPC has replaced (#70). Fargate retires connector tasks for platform patching every few days, and every upgrade that changes the connector task definition re-registers the connector; the old node stayed in the stale count, and the alarm with it, until the 24-hour prune. The reconciler now counts a node as stale only when its account and VPC has no online trusted connector, lists a replaced node assupersededConnectors(the EMF log line, thereconciledlog and the site's status row) and still prunes it 24 hours after it went offline. The alarm's metric, threshold and period are unchanged; its description names the new rule. Upgrade:NetworkStack(manual,--exclusively). -
Private network: a user's first connect to a site reaches their granted resources within seconds (#76). The reconciler leaves out users Headscale does not know yet, and the device minter creates that user on the first enrollment, so the rules waited for the next scheduled pass (up to 5 minutes). The device minter now starts a pass (an async invoke of
<prefix>-acl-reconciler, named by its newNETWORK_RECONCILER_FN_NAMEvariable) whenever it creates a Headscale user; a failed start is logged as a warning and the key is still returned. Later mints start nothing. Upgrade:NetworkStack(manual,--exclusively). -
Private network: Security Hub ELB.17 passes again for the Headscale ALB listener (#78). AWS removed
ELBSecurityPolicy-TLS13-1-2-Res-2021-06from the control's accepted policies, so existing deployments started failing it with no change on our side. The listener now usesELBSecurityPolicy-TLS13-1-2-Res-PQ-2025-09: the same TLS 1.2/1.3 protocols and ciphers plus hybrid post-quantum key exchange, so no client loses access; the listener is updated in place. Upgrade:NetworkStack(manual,--exclusively). -
The credential issuer's KMS signing keys carry an alias per key version,
alias/<prefix>-oidc-signing-v<N>, so the KMS console names them (#76). The keys and their policies are unchanged. Upgrade:CredsStack. -
Desktop (Windows): the tray row Update … available · Open download page opens
https://github.com/FNTIO/aws-infrastructure-client/releases/latest; since 0.5.0 it openedgithub.com/undefined/undefined/…, because the packagedpackage.jsonhas nobuildblock to read the feed from (#69). macOS, which downloads and installs from the tray, was not affected. -
Desktop: a saved deployment that is waiting for the network or unreachable shows its status as a chip beside its name (no network in amber, unreachable in red) instead of a suffix on the name; the name ellipsizes, the chip never does, so the status stays readable for any name length (#72). Hover text, click-to-retry and ✕ are unchanged.
-
Desktop: About Infra Client and Check for Updates… in the app menu (#71). On macOS the check sits below About in the app menu, which keeps every default entry; Windows and Linux get a Help menu with both, so Windows shows the installed version for the first time. The tray has the check too while no update is pending. A manual check always answers: the latest version, a restart on macOS once the update is downloaded, the Release page on Windows, or the error. The automatic checks stay silent, as before.
Full changelog: CHANGELOG.md · npm: @fntio/infra-client
v0.6.0
Upgrade: hub stacks (DataStack CredsStack WebStack), then NetworkStack if you run the private network (the device list and delete routes are answered by the device minter). If an admin had set the dashboard layout from the admin Accounts page, put that value in the config (dashboard: { layout: "sections" | "plain" }) before deploying: the stored value is no longer read and the deployment runs in auto until the block is set.
- Personal access tokens (#52), opt-in with
personalAccessTokens: { enabled: true, maxLifetimeDays? }: every user creates tokens for scripts, CI jobs, command-line tools and remote development machines on a new Access tokens page (user menu) — a name, an optional description, scopes (api,aws,network,shares), an expiry up tomaxLifetimeDays(default 90) and, foraws/api, an optional subset of their roles. The secret (icpat_<tokenId>_<secret>) is shown once; only its SHA-256 is stored, in a newTokensTableinDataStack. A token acts as its owner with the owner's current access narrowed to its scopes and role subset (decided inaccess.ts);apireaches every route the owner could, admin routes only while the owner is an admin; a route outside the scopes answers403withcode: "TOKEN_SCOPE". Tokens cannot create, list or revoke tokens, and plugin routes refuse them. A disabled or deleted owner's tokens stop working; deleting a user deletes their tokens; revoking takes effect on the next request. New routesGET /tokens/current,GET/POST /tokens,DELETE /tokens/{id},GET /tokens/expiry,POST /tokens/expiry/seen,GET /admin/tokens,DELETE /admin/tokens/{id}; the admin Users page gains an Access tokens tab; the bell warns before a token expires and once after. Activity rows written through a token carryactor.token({ id, name }). Discovery gainspersonalAccessTokens.enabled. - Service users carry scopes (#48): an admin picks
aws,network,shares,admin.read.alloradmin.write.allwhen creating one (required onPOST /admin/service-users) and changes them later withPUT /admin/service-users/{clientId}/scopes. The narrow three share the personal-token route table;admin.read.allopens theGETroutes under/adminandadmin.write.allevery/adminroute (it implies the reads). Scopes are read from the registry row on every request, so a change holds from the service user's next call; a route outside them answers403withcode: "TOKEN_SCOPE". Only the admin scopes make a service user an app admin, for the methods they cover; a service user without them vends the roles granted to it and enrolls machines on the private network (POST /network/device-key,ephemeral: true) with nothing else, and anadmin.write.allone manages users and service users like a human admin. Breaking:POST /admin/service-usersrequiresscopes, and every service user is now route-gated. A service user created before this release reads asaws,network,shares,admin.write.all, which reaches the/admin, vending, device-key and share routes; routes outside that table (/requests,/notifications,/preferences,/resources,/plugins) answer403for every service user, so an integration that used them needs a person's session or a personal access token. The Users page's Service users tab gains scope checkboxes on create, scope badges per row, a Change scopes editor and a Network resources section on the panel (service users were already valid resource-grant principals through the API). Activity gainsserviceuser.scopes. The wiki's service-user runbook carries the headless vend-and-join sequence for a VM agent. Personal access tokens are unchanged. - The tenant / project / stage filter popup beside an Account picker (Request a role, the group and role forms) opens leftwards from its button instead of running off the right edge of the panel (#66).
- Dark mode: the Account picker’s dropdown list (Request a role, the admin forms) sits on the solid popover surface like every other floating panel instead of a translucent overlay the page showed through (#57).
- The dashboard layout is deployment config (#56): a new optional
dashboard: { layout: "auto" | "sections" | "plain" }block (defaultauto, validated at synth) reaches the API asDASHBOARD_LAYOUT; the Dashboard select on the admin Accounts page andGET/PUT /admin/dashboard-layoutare gone, and the__dashboard-layout__row in the Resources table is no longer read or written.GET /accounts/homekeepslayoutandsectioned. - Desktop: a saved deployment that nobody could reach at launch (the app reopening after a reboot before Wi-Fi is back, a power-loss restart) is no longer marked unreachable for good: its tab reads · waiting for network and the app re-probes it on its own (every 10 s while offline, then with backoff for about ten minutes once online) before it settles as · unreachable; a host that answers but is not an Infra deployment is unreachable at once. Clicking a waiting or unreachable tab retries it (#58).
- Private network:
POST /network/device-keytakesephemeral: true— the device is removed by the control plane 30 minutes after it goes offline instead of being expired after 7 idle days.GET /network/deviceslists the caller's own devices on every site andDELETE /network/devices/{siteId}/{name}removes one (404 for anyone else's); the Network page lists them under My devices.GET /networkgainsdevShares: { enabled, maxDurationHours }. - Dev shares: a share can be bound to one device (
deviceonPOST /network/shares, the device's name on the site). The port then opens on that device only — resolved by name on every reconciler pass, so a device that re-enrolls under the same name keeps its shares — and always to the owner's own other devices, so a bound share may name no grantees. The one-share-per-port rule applies per device. AdurationHoursabovemaxDurationHoursis now clamped to it instead of refused. In the desktop app the share form gains a Serve from picker when you have other devices on the site, the share card names the device, and a device's relay leaves shares bound to another device alone. - Dev shares:
network.devShares.maxActiveSharesPerOwner(1–200, default 10, checked at synth) sets how many live shares one user may hold — the fixed limit of 10 before; raise it where tools share ports of many machines for one person.GET /networkreports it next tomaxDurationHours. - Threat protection (
loginMethods.threatProtection): a Medium-risk sign-in — in practice a known user on a new network — is now allowed instead of "MFA if configured". Cognito offers only the password on any attempt it applies MFA to, so for a user without TOTP that action removed passkey and email code without adding a second factor. High risk keeps the step-up and, whenemailis configured, emails the user about the attempt. Upgrade:CredsStack. - Sign-in sends Cognito a device fingerprint (
UserContextData, AWS's collector vendored into the web app) on everyInitiateAuthandRespondToAuthChallenge, so threat protection scores a known device on a new network lower than a stranger. When a user's usual passkey or email-code sign-in is missing from the chooser because an MFA action limited the attempt to the password, the page says so in one line. GET /accounts: every role carriessessionDurationSeconds(how long its credentials last).POST /credentialsalways returnsexpiration.apiThrottle: { rateLimit?, burstLimit? }(default 100 rps, burst 200) sets the vending API's stage throttle, which was a fixed 25 rps with a burst of 50 since July. That ceiling applies across every caller — one page load is 10 to 15 calls, so a burst of 50 was three or four people opening the dashboard at once, and the e2e suite tripped it on its own. The per-token WAF rule (waf.perTokenRateLimit) is what bounds one leaked token. Upgrade:CredsStack.- Review fixes before release:
DELETE /network/devices/{siteId}/{name}removes exactly the device listed under that name — a stale duplicate no longer takes the re-enrolled machine (same hostname, suffixed name) with it.GET /network/devicesanswers for the sites that are up when one control plane is down and names the others inunreachable(My devices says so). A device-bound share refuses ahostnamethat is not the bound device's. The per-user token cap holds under parallel creates, andPOST /tokens/expiry/seencounts only the notices it flagged. - Network page: one block per VPC whatever the number of sites — the bar names the VPC, its account and (with several sites) its site as a chip, and carries that site's Connect / Switch & connect; the Shares group and My devices follow. Section chips under the header (one per VPC, Shares, My devices, with counts) scroll to each block. Multi-site deployments no longer nest VPCs under a block per network.
Full changelog: CHANGELOG.md · npm: @fntio/infra-client
v0.5.0
Upgrade: hub stacks (DataStack CredsStack WebStack), then NetworkStack and every Connector-<account> stack if you run the private network — see the wiki's Deploy and Upgrade page, "Upgrading to 0.5.0".
- Bell notices for a change that moves many targets at once are one row, not one per target: when one person gains or loses more than five roles, resources or plugin permissions through a group (a membership or bundle edit) or a cascade, the feed gets a single
kind: BUNDLEnotice carryingcountand, for a group,groupId/groupName("7,346 access grants · group · FullSystemAccess"). Adding a member to a 7,000-role group used to write 7,000 "access granted" rows for them, the pop-up logic re-raised that backlog as a fresh summary every minute, andPOST /notifications/revocations/seen(one update at a time) could not clear it inside the request timeout. Mark-seen now runs 25 updates at a time; the web app folds raised notice keys older than its 500-key cap into a watermark instead of forgetting them. The activity log's per-person rows under such a change collapse the same way: oneaccess.gain/access.loserow per person carryingdetails.count("gained 7346 access grants via group FullSystemAccess") instead of one per target; the primary row'saffectedlist is unchanged. - Renew sign-in before the 12-hour wall, from anywhere: the user menu's "Session ends" row becomes a "Renew sign-in" button that shows the time left (the end time is its tooltip); the desktop tray's session row gains "Renew sign-in…" (desktop app 0.5.0 or later). All of them start a fresh sign-in while the running session, its credentials and the VPN stay up until the new one lands — nothing is torn down mid-task.
- Desktop: closing a backend tab no longer loses its address. The connect screen (the + button) lists every deployment the app has connected to at least once under Recently connected, newest first with its project name; click one to reconnect, or its ✕ to forget it after a confirmation. Only that ✕ removes an entry (
recents.jsonin the app's data folder, 20 entries). Reconnecting a saved deployment that was unreachable at startup now keeps its login session instead of starting a fresh one. - Large catalogs (#39): the portal now holds a 4,000-account, 16,000-role catalog without the pages or the API growing with it. The catalog (accounts, roles, group names) lives in an in-memory index per API container, rebuilt only when a catalog write moves a version row, and every catalog-sized list — my access, the admin Accounts page, the request picker, the users directory, a group's bundle — serves one paged contract (
?q=&filter[key]=a,b&sort=&limit=&cursor=→{items,nextCursor,total,facets}; withoutlimiteach route answers its old whole-list shape, so existing clients keep working). Access is still decided in one place before any page is cut; the index lists, it never grants. - Dashboard (#45): a user who holds an account an admin kept off the dashboard gets a sticky search over everything they hold — results drop down as role rows with the card's actions (star, Console, Get credentials, the desktop auto-login bolt) — above Signed in now (roles with running credentials), Main, Favorites and Recently vended, each heading pinned while its cards scroll, and a table of the Other accounts (the roles outside Main and Favorites, five to start and more as you scroll;
GET /accountsgains themainandfavoritefacets and a sparsemain: falseon each account for it); everyone else keeps the plain page unchanged. - Lists in panels and dialogs (a user's or group's roles, a service user's roles, a group editor's picked roles) show five rows, with Show more / Show less and a search box as soon as any are folded away, instead of thirty and "Show more"; the grant-roles dialog is wider and its dimension filter no longer opens off the dialog's edge. New account flag Show in the Main section (on by default;
main: falseonPOST/PATCH /admin/accounts),GET /accounts/home,PUT /preferences/favorites/{account}(50 favorites, synced across devices), recents from credential and console vends. The admin Accounts page has a deployment-wide Dashboard layout choice (PUT /admin/dashboard-layout): automatic (the rule above), sections for everyone, or the plain list for everyone. - Account dimensions (#47), opt-in:
accountDimensionsin the project config stores tenant / project / stage on each account when it is registered, read off its name (<tenant>-<project>-<stage>; a pipeline may passdimsinstead), admin-editable on the account form and never re-derived by a rename. They filter the dashboard search, the admin Accounts page, every account picker (request form, role form, grant dialogs) and the group editor, which gains add one role across many accounts (POST /admin/groups/{id}/roles/bulk, with a dry-run count). The role card's silhouette follows the stage through the config's stage map (stages); without the block nothing changes. A deployment turning it on backfills existing rows withscripts/backfill-dimensions.mjs(deployment README "Account dimensions"). The dashboard's search and browse results can be grouped by one dimension (a Group by choice beside the filter), and registering a role offers the groups that already bundle that role in the account's sibling accounts (same tenant and stage,GET /admin/groups/suggest). - Groups at scale (#43, #44):
POST /admin/groups/{id}/rolesandDELETE …/roles/{account}/{role}add or remove one role, written only if absent, so a pipeline registering accounts never races another writer (the full-listPATCHis read-modify-write); the add answers with the change ({groupId, name, role, added}), not the bundle, and neither call reads the whole bundle — a 4,000-role group answers in well under a second; the access diff behind a bundle edit reads only the touched targets; the group list caps each bundle at 50 roles withroleCountandGET /admin/groups/{id}/rolespages the rest; one yearly review per group (its clock = the latest approval or bundle change) replaces one per bundled role; the users directory caps each user's role and resource lists at 50 with totals,GET /admin/users/{id}carries everything. - Registration is idempotent (#43):
POST /admin/accountsandPOST /admin/accounts/{a}/rolesanswer200with the stored row when the same registration is sent again,409only when a field differs — a pipeline can retry any step. The admin forms still refuse a duplicate. - Target-account template (#43):
TargetAccountStack/targetAccountTemplate()synthesize one account-agnostic CloudFormation template (nativeAWS::IAM::OIDCProvider, sub pin throughAWS::AccountId) for StackSets or Control Tower, so a 4,000-account deployment does not need one CDK stack per account. - Pickers (#46): every place that listed the whole catalog in a
<select>or a checkbox wall (request form, add-role form, group form, the grant dialogs on user, group and service-user panels) is a typeahead over one page at a time. - Access report (#46):
POST /admin/access-report/jobsbuilds the CSV in the background (a second Lambda on the API bundle, 15-minute budget, private reports bucket with a 7-day expiry) andGET …/jobs/{id}hands back a 15-minute download link; the Users page's export uses it. The inlineGET /admin/access-reportremains. - Reads bounded by the catalog (#41, #44):
BatchGetItemretries unprocessed keys and runs in parallel chunks; the admin bell reads the bare directory; the setup checklist counts from the index; the reconciler scans only what it needs. - Security Hub S3.9 and CloudFront.5: every hub bucket (web app, OIDC issuer, plugin assets) sends S3 server-access logs, and both CloudFront distributions send standard logs, into one new
AccessLogsBucketinDataStack(90-day expiry, kept on teardown whenprodis on). That bucket is the one that logs nowhere; AWS's own S3.9 text says to suppress its finding. Rides a deploy of all three stacks. - Private network: the ALB writes its access logs to a bucket of its own in
NetworkStack(Security Hub ELB.5): SSE-S3, TLS-only, public access blocked, bucket-owner-enforced, objects expire after 90 days, its own access logs go to the hub's shared access-log bucket unders3/alb-logs/; the bucket is kept on aprodteardown and emptied and deleted otherwise. Rides aNetworkStackdeploy. - Private network: ECS.20 passes on every network task definition. The control-server task is three non-root containers with no shared task volume: the init container renders the config and the self-signed TLS pair next to the database on the EFS mount (the access point performs every file operation as uid 0, whatever user the container runs as), the server container serves from there with its Unix socket on a per-container tmpfs and runs the two CLI bootstrap steps itself once the socket answers, so the
bootstrap-userandbootstrap-keysidecars are gone andCustom::HeadscaleApiKeyreads the key from theheadscalecontainer's own log stream (headscale-<site>/headscale/<task>); the publish-cert step reads the certificate from EFS. The connector task's two containers are non-root too, each with its own tmpfs, and the minted auth key crosses from the aws-cli container to the Tailscale container over the task's localhost instead of a shared file. The control server moves to the upstream-debugimage tag (same Headscale, plus busybox for the start script). Why the layout changed: a Fargate task volume is root-owned 0755 when the task starts whatever mode the image gives the path, so 0.4.0's non-root containers could only share a volume a root init container had chowned, which kept ECS.20 failing on that container and on the connectors. Rides aNetworkStackdeploy (each control server restarts once; the EFS directory gainsconfig.yaml,tls.keyand `tls...