- rejects a remove or update request whose terminator is owned by a different
router, closing the two fabric handlers the batch fix did not cover
- adds a unit test for the single-terminator ownership check
- adds an end-to-end test that drives the fabric control channel from an
enrolled router against a second router's terminator, covering single remove,
batch remove, and re-weight, plus a control that a router can still remove its
own
- verifies the control-channel peer leaf against the controller's full trusted-CA pool
(identity.CA()) instead of only self-signed roots, honoring intermediate trust anchors
and multi-root bundles
- drops the client-auth extended-key-usage requirement so an externally managed PKI with
arbitrary or absent EKUs is not rejected
- adds tests asserting router control channel connections with an untrusted or
self-signed leaf (including one backed by a scraped CA-chained filler cert) are
rejected, and that separately-validated channel types are skipped
- extracts a small header helper so the grouped-connection first-underlay scoping
is unit-testable, and tests that the flag is not inherited by additional underlays
Runs the router certificate fingerprint validation for router control-channel
underlay types, which was previously skipped for any connection carrying a
channel type header. Connections of other types (e.g. the raft mesh) continue to
be deferred to their own acceptor.
The already-connected / churn guard is applied only when establishing a new
channel, so additional underlays of a grouped control channel are not rejected
while the router is already connected.
- filters the fabric RemoveTerminators handler so a router can only remove terminators
whose owning router matches the request source; ids owned by another router are dropped
and logged rather than deleted
- keeps absent ids so a delete racing a not-yet-applied create is still ordered after it
- the edge control channel already enforces this via verifyTerminator
Backport of #4165. release-v2.0.x has no RemoveTerminatorsV2, so only the v1 handler
is affected.
Backport of #4160 to release-v2.0.x.
- adds command.WasLeaderless to classify cluster-has-no-leader dispatch errors as retriable
- replies busy instead of dropping or hard-failing terminator creates when the cluster is briefly leaderless, so the router backs off and requeues promptly rather than waiting for its multi-minute recovery scan
- removes the racy up-front leaderless pre-check in the sdk create handler in favor of classifying the actual dispatch result
- applies the same retriable classification to the ert tunnel create and batch remove terminator handlers
- changes lastResponse from plain int64 to atomic.Int64 in both router
and peer heartbeat callbacks, fixing a data race between the heartbeat
response handler and the heartbeat check ticker
- fixes peer heartbeat logger channelType from "router" to "peer"
- adds CreateCircuitV3 message type and handler for routers that have
already authorized dials locally via RDM, bypassing service session
tokens in favor of identity ID, service ID, and pre-assigned circuit ID
- renames CreateCircuitRequest/Response to CreateCircuitV2Request/V2Response
for clarity now that V3 exists
- adds CircuitManager.Reserve to atomically claim circuit IDs before routing,
preventing collisions on pre-assigned IDs
- extends CreateCircuitParams with GetCircuitId so V3 can supply a
pre-assigned circuit ID (falls back to UUID generation when empty)
- fixes IsDialableByIdentity which was incorrectly calling IsBindableByIdentity
- extracts V2 handler into its own file create_circuit_v2.go
- adds CreateCircuitV3RequestType/ResponseType (20222/20223) to edge_ctrl protobuf
- registers V3 handler in controller server
- removes legacy v1 terminator code path; all terminators now use v2 flow
- refactors edgeTerminator.close() to decouple SDK notification from control plane notification
- adds pending SDK close notification queue with retry when channel is busy
- adds post-create inspect mechanism that verifies SDK still holds the bind after terminator creation
- queues second post-create inspect when establishment takes >30s to catch SDK timeout races
- detects and discards stale reordered binds on the same connection by comparing connIds
- re-establishes replacement terminators when a delete/create race is detected
- eliminates IsEntityPresent pre-filter in removeTerminatorsHandler to prevent raft ordering races
- fixes ValidateTerminators to query identities from the correct manager with the correct filter field
- adds postCreate flag to ValidateTerminatorsV2Request so routers skip redundant SDK inspect
- returns retry-later (nil result) from router validation when inspect is temporarily unavailable
- blocks SyncAllSubscribers until completion and guards RouterDataModel replacement with in-progress flag
- fixes InheritLocalData to enable service access tracking for all subscribed identities
- adds Services.Has check in GetServiceAccessPolicies to prevent false policy grants
- validates policy-to-identity associations in ValidateServicePolicies
- adds `ziti agent tunnel dump-sdk` command for SDK context inspection via IPC agent
- adds `ziti fabric inspect sdk` command to query SDK context through routers
- fixes --expected-per-host CLI flag binding in validate terminators command
- changes bind-access-lost retry hint from NotRetriable to RetryStartOver
- moves trace route response and xgress close handling off channel handler goroutine
- fixes listTerminators test helper to URL-encode filter parameter
- improves sdk-hosting-test validation resilience with login and query retries
- adds terminator_create_flow.md documenting the full lifecycle across SDK, router, and controller
- adds detailed logging for data model sync, service access tracking, and subscriber change detection
* Allow routers to request current cluster membership information. Fixes#3503
* Get cluster membership information from raft directly, rather than trying to cache it in the DB. Fixes#3501
* Set a router data model timeline when initializing a new HA setup, rather than letting it stay blank. Fixes#3500
* Reduce router data model full state updates. Fixes#3504
* When routers change (connect/disconnect), we re-check the mesh. We are passing in the routers to the channel and they can build up, which is unnecessary. We only need to signal once for all the routers that change since the last time we checked, that the mesh needs to be checked.
* We pass link changes through the run method, which is unnecessary since we pass it to a new goroutine. This inefficiency should be fixed.
* We're evaluating all faulted links for rerouting, even if they were still pending. Only reroute connected links.
* We're using time.After in each for loop, which can accumulate timers. Use a single Ticker instead.
Paul Lorenz