acc: run ssh/connection locally#5907
Conversation
Flip acceptance/ssh/connection to Local = true so it runs against the in-process testserver (it was disabled on both cloud and local in #4838). This is the dedicated-cluster counterpart to connect-serverless-gpu: a full SSH handshake can't complete locally (no real compute/sshd + a wss driver-proxy), so the local run drives the connect flow up to the connection attempt and asserts the dedicated-cluster bootstrap job the CLI submits (existing_cluster_id, secret scope, notebook task, serverless=false). Make the default test cluster a running dedicated single-user cluster so `ssh connect --cluster` passes ValidateClusterAccess and the cluster-state check, matching the cloud TEST_DEFAULT_CLUSTER_ID this fixture stands in for. The connect attempt itself is expected to fail locally; its noisy output goes to LOG.stderr (excluded from the golden files).
Integration test reportCommit: 1fc5b95
8 interesting tests: 4 SKIP, 3 KNOWN, 1 RECOVERED
Top 10 slowest tests (at least 2 minutes):
|
anton-107
left a comment
There was a problem hiding this comment.
Approve. Clean and low-risk. Confirmed TestAccept/ssh/connection runs and passes locally (~0.9s) on the Linux job.
I verified the shared-fixture change has no golden blast radius: the only acceptance test that reads TestDefaultClusterId by GET is bundle/deployment/bind/cluster, which projects with jq "{cluster_name}", so the new State/DataSecurityMode/SingleUserName fields are filtered out and no other golden changes — matching the PR description. The change is also semantically more faithful to the cloud TEST_DEFAULT_CLUSTER_ID it stands in for.
The local run is a change-detector for the submitted bootstrap-job payload (the connect attempt is expected to fail; its output is correctly routed to LOG.stderr and excluded from goldens), which is a reasonable and deterministic thing to assert. Nice follow-up once #5920 lands: its websocket-scheme fix + /ssh handler would let this dedicated-cluster case complete a real handshake locally too.
Changes
The dedicated-cluster counterpart to
acceptance/ssh/connect-serverless-gpu(#5878). Flipsacceptance/ssh/connectiontoLocal = trueand asserts the bootstrap job the CLI submits forssh connect --cluster(existing_cluster_id, secret scope, notebook task,serverless=false), reusing the testserver SSH-tunnel stubs from #5878.The default test cluster (
TestDefaultClusterId) is now a running dedicated single-user cluster — the shapessh connect --clusterrequires (ValidateClusterAccessrejects anything else), matching the cloudTEST_DEFAULT_CLUSTER_IDit stands in for. Onlybundle/deployment/bind/cluster(filters tocluster_name) and thecmd/apiroute stubs reference it, so no other goldens change.Tests
acceptance/ssh/connection(asserts the submitted dedicated-cluster bootstrap job payload)