Commit Graph

7 Commits

Author SHA1 Message Date
Joseph Doherty dc0a24f224 feat(galaxy): operator-authenticated alarm ack — send Galaxy credential to mxaccessgw (#749)
v2-ci / unit-tests (push) Failing after 1h8m43s
mxaccessgw 488649c+ acknowledges by AuthenticateUser as the operator and
writing <alarm>.AckMsg, so AcknowledgeAlarmRequest now needs a real Galaxy
operator_user + operator_domain + operator_password. The OPC UA data plane is
API-key/anonymous and carries no Galaxy password, so the driver resolves the
account per ack:

1. the session's API key's Galaxy ack identity, set on AdminUI /apikeys and
   held ONLY in the encrypted Secrets store
   (otopcua/apikeys/<keyId>/galaxy-ack-{user,password}); a key with a user but
   no password secret is refused, never silently downgraded;
2. else the driver instance's AlarmAck fallback account (DriverConfig
   FallbackUser/FallbackDomain/FallbackPasswordSecretRef; secret:/${secret:}/
   env:/file:/dev: only — a literal is refused);
3. else DriverOperationRefusedException: nothing is sent, never an empty
   operator_user. Refusals are excluded from the resilience retry/breaker.

- Pins ZB.MOM.WW.MxGateway.Client/Contracts 0.2.1 -> 0.3.0 (vendored into
  nuget-packages/ from the Gitea feed, per the local-mxgw source mapping).
- The session ApiKeyId (not a secret) rides NativeAlarmAck ->
  RouteNativeAlarmAck -> RouteAlarmAck -> AlarmAcknowledgeRequest; the password
  is resolved in the driver and never travels through actor messages.
- The OPC UA principal + key id are appended to the AckMsg comment
  ("[OPC UA: <principal>, key <id>]") and logged, so a shared fallback
  account does not lose who acknowledged.
- Password never logged: GalaxyAckCredential.ToString omits it, the
  acknowledger scrubs it from every gateway message it echoes and withholds
  any exception that echoes it; audit rows carry only the key id.
- AdminUI: /apikeys "Ack identity" editor (set/remove; delete-key clears it),
  Galaxy driver form gains the fallback section.
- Synchronous OPC UA status for a refusal is follow-up #760 (the native ack
  path is fire-and-forget after the Part 9 method returns).

Tests: request mapping (password populated), password absent from logs and
exception text (mutation-checked), key-mapped / fallback / refused paths at
resolver and driver level, breaker non-trip, key-id propagation through node
manager and DriverHostActor, AdminUI service + form round-trip.

Closes #749
2026-11-06 16:23:00 -05:00
Joseph Doherty c9fdcc3ea6 fix(adminui): clear stale one-time key banner on delete; modal backdrop no longer overruns viewport (#631)
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
2026-10-08 14:21:20 -04:00
Joseph Doherty 1475991fd6 fix(adminui,client): error boundary inside the interactive scope, stale-commit OnSaved, AbCip browse delay emit, DisposeBridge cancel (review R3)
M1: PageErrorBoundary wraps the body of every InteractiveServer page; the inert layout boundary is removed.
m1: DisposeBridge cancels its CTS when the 5 s wait times out.
m2: a CSV import that commits after the modal closed still raises OnSaved.
m3: document SecurityMode=None no-fallback in Client.CLI.md / Client.UI.md.
m4: browsePageDelayMs written only when loaded or changed; negative flagged.

Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
2026-10-08 07:31:19 -04:00
Joseph Doherty d3bca7afd4 fix(adminui): stale-open guards, validation summaries, key-delete confirm, error boundary (#604)
Claude-Session: https://claude.ai/code/session_01Ef2ozxS7QPanGT8uoATbTe
2026-10-08 06:55:23 -04:00
Joseph Doherty 3d6d28e5a2 style(adminui): btn-groups + prose trim on top-level ops pages
Group related action buttons into btn-group/btn-group-sm on Alerts, Certificates,
ApiKeys and ScriptEdit; promote the RoleGrants Delete from btn-link to
btn-outline-danger. Replace the two inline style="border-color:var(--alert)"
no-ops with the .notice-alert class, and the undefined --surface-2 token in
Scripts with the theme's shipped --zebra-bg.

Delete unconditional documentation prose now covered by docs/ (Redundancy,
Telemetry, AlarmTracking, security, Reservations, AlarmHistorian, ScriptEditor),
keeping state banners and live captions. Correct the LDAP config key on
RoleGrants to Security:Ldap:GroupToRole (LdapOptions.SectionName) and drop the
same wrong key from Account with its deleted block; reword the stale
"Engine emit (F8/F9) is pending" script-log empty state.
2026-08-10 16:32:29 -04:00
Joseph Doherty c1771dc271 docs(acl): correct the API-key revocation warning after live verification (#520)
The /apikeys page claimed revocation "takes effect on a node only after the next
deployment". The live gate showed that is wrong for the node the operator is
most likely looking at: revoking through the UI denied the key on the VERY NEXT
request against the fused admin+driver node, because ConfigDbApiKeyStore reads
per request. The deployment delay applies only to driver-only nodes, whose keys
ride the artifact.

Reworded to say both, and to tell an operator to assume the slower behaviour when
revoking a compromised key — which is the decision the warning exists to inform.

Found by doing the thing rather than reasoning about it: the previous wording
looked obviously correct until the revoked key was actually re-presented.

Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
2026-07-31 14:06:13 -04:00
Joseph Doherty 087a969ee6 feat(acl): API key minting — AdminUI page + admin service (#520)
Pulled forward from stage 7 because without it the feature is unusable and
un-live-gateable: there was no way to create a key short of hand-inserting a row
and computing an HMAC by hand.

The minting logic is an IApiKeyAdminService in Core.Authorization, NOT logic in
the Razor page — this repo has no bUnit, so anything only a page can reach is
untestable, and minting credentials is not where that belongs. The page is a thin
shell over it.

Moved to Core.Authorization so both ends can reach them: ApiKeyOptions (was in
Host) and OpcUaPrincipals (was in OpcUaServer.Security). The OPC UA server
resolves principals; the AdminUI must refuse to mint a key claiming a reserved
one. OpcUaServer already gained a Core reference in stage 1b, so no new edge —
AdminUI gains one.

ApiKeyOptions is now bound UNCONDITIONALLY rather than in the driver branch: a
driver node needs the pepper to VERIFY keys and an admin node needs it to MINT
them, and getting that wrong on either side is a silent authentication failure
rather than a startup error.

Validation rules, all covered: the reserved 'anonymous' principal is refused
case-insensitively (so "ANONYMOUS" cannot slip past into a distinct principal
that the evaluator would then match case-INsensitively against anonymous's
grants); blank name/principal; a principal longer than the column, rejected
rather than truncated, because silent truncation yields a key that authenticates
and is granted nothing with no visible cause; a past expiry. Generation runs
BEFORE the DbContext is opened so an unconfigured pepper surfaces as a clear
error instead of a stored row whose hash nothing can verify.

The page shows the plaintext ONCE and says so; it is held in component state and
written nowhere else, because the server keeps only a hash. Revocation is flagged
in the UI as taking effect only after the next deployment — driver-only nodes
have no ConfigDb, so keys ride the artifact.

FleetAdmin-only (Administrator), not ConfigEditor: this mints credentials.

Core.Tests 68 green.

Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
2026-07-31 13:40:34 -04:00