* [test] fix TestProcessPriority
Need to give ws-daemon time before it applies process priority
* [server] always limit network connections
We are not limiting network connections for preview environments
* Cleanup
* Fix
* Always do network limiting
* Regardless of billing mode
* If there's an error
* Wait a bit longer, just in-case, to make the test more durable
* Cleanup
* No parallel
* [preview] update to use gen111 image
* [test] fix TestOpenWorkspaceFromPrebuild test
It was failing with an exit code of 2 on grep, because it was trying to grep a folder, rather than recurse into it.
When the grep command was run manually from a workspace that started from a prebuild, we saw `grep: logs: Is a directory`. So, it's important for integration tests to output the exit code, but also error text.
* [test] fix TestOpenWorkspaceFromPrebuild
We need to wait for the started workspace to stop, before we try to restart it
* be parallel again
* Nope
* Set StripeSecret on Preview install
Otherwise this returns false:
643701c635/components/server/src/billing/billing-mode.ts (L25)
Because EnablePayment is set to false at install time:
643701c635/install/installer/pkg/components/server/configmap.go (L282)
* [preview] update to gen108 VM image
* [preview] sleep longer
This is a bandaid. With a shorter sleep, the API server tries to come online, but fails. As a result, core-dns stays pending, and calico and other manifests are never applied.
* [preview] no local storage
* [preview] wait for containerd
* [preview] disable-network-policy, as per k3s suggestion when using calico as CNI
ref: https://docs.k3s.io/installation/network-options#custom-cni
* [preview] improve parity with prod
* [preview] gain more parity with prod
* [preview] scale coredns to 1
* [preview] use out of the box coredns
* remove tweak for containerd readiness in cloudinit.yaml
* [docker-up] update docker compose to 2.18.1-gitpod.2
* [workspace] update containerd
From v1.6.20 -> v1.6.21
* [image-builder-bob] fix build error after containerd update
Context:
[components/image-builder-bob:app] level=error msg="Running error: 1 error occurred:\n\t* can't run linter goanalysis_metalinter: buildir: failed to load package tracetransform: could not load export data: no export data for \"go.opentelemetry.io/otel/exporters/otlp/otlptrace/internal/tracetransform\"\n\n"
* [preview] update VM image with fewer cached images
This will help us avoid pod eviction and improve stability
Related: https://github.com/gitpod-io/gitpod-packer-gcp-image/pull/235
* [test] more cowbell
* [tests] unique ContextURL for commit only
* [tests] include std output in the error message for commit
Sometimes commit fails with status 1 w/o error output...so it's helpful to look at StdOut
* [tests] Fix TestGitActions by using the test case context
Prior to this, createWorkspace was working, but getWorkspace could not find the created workspace.
* [test] bypass exit code 1 on `git commit` with `--allow-empty`
* [test] Fix gcloud auth, do setup before auth
* Show all output
Might remove later...
* [test] avoid using deleted user, identity, and token
* [test] add organizationId to CreateWorkspaceOptions
It's expected on the Typescript side:
90b7d65858/components/server/src/workspace/gitpod-server-impl.ts (L1227-L1235)
* [test] orgId is required on createWorkspace
But sometimes there's no team 🤷
* [test] fix git context tests
We use UBP now, there is no more unleashed.
Also, remove the "ff" feature flag code (which was for PVC). It was mutating the username, resulting in Code 460 errors on createWorkspace
* [test] Use example test as the example
* [test] fix context tests when run as gitpod-integration-test user
* [test] clean-up
* [test] wait for workspaces to stop
Tests intermittently fail with to avoid intermittent failures
* [test] add code owners
This way, we can assert tests are passing for all teams prior to merging
* [test] limit # of tests that can run in parallel
* [test] no parallel tests
Test to see if flakeyness goes away...
...and bump the timeout because we reduced parallel runs
* [preview] update the VM image to have parity with production
This:
1. updates from K3s 1.23 to 1.26
2. requires that we remove PodSecurityPolicy changes (as it's no longer supported)
3. resolves intermittent disk pressure issues
* [preview] no PSP in support of VM image update
* We were getting PSP from rook/ceph, which I think was for PVC
* We were getting PSP from the monitoring-satellite
* [test] don't wait for workspace stop with git_test.go, we're testing git actions.
Why? We miss state transitions, it's not guaranteed each one will be returned, and there are other tests waiting.
For example, in the below log, we miss INITIALIZING, RUNNING, and STOPPING.
workspace.go:369: attempt to create the workspace as user 0565bb3c-e724-4da9-84fb-22e2a7b23b8c, with context github.com/gitpod-io/gitpod-test-repo/tree/integration-test/commit
workspace.go:411: attempt to get the workspace information: gitpodio-gitpodtestrepo-nscsowy1njb
workspace.go:423: not preparing
workspace.go:432: got the workspace information: gitpodio-gitpodtestrepo-nscsowy1njb
workspace.go:460: wait for workspace to be fully up and running
workspace.go:569: prepare for a connection with ws-manager
workspace.go:590: established for a connection with ws-manager
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:631: status: 462f1325-3019-4547-8666-508e8353335e, PENDING
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:631: status: 462f1325-3019-4547-8666-508e8353335e, PENDING
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:631: status: 462f1325-3019-4547-8666-508e8353335e, CREATING
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:631: status: 462f1325-3019-4547-8666-508e8353335e, CREATING
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:631: status: 462f1325-3019-4547-8666-508e8353335e, CREATING
workspace.go:598: check if the status of workspace is in the running phase: 462f1325-3019-4547-8666-508e8353335e
workspace.go:504: waiting for stopping the workspace: 462f1325-3019-4547-8666-508e8353335e
workspace.go:514: attemp to delete the workspace: 462f1325-3019-4547-8666-508e8353335e
workspace.go:797: confirmed the worksapce is stopped: 462f1325-3019-4547-8666-508e8353335e, STOPPED
workspace.go:538: successfully terminated workspace
git_test.go:172: failed to wait for the workspace to start up: cannot wait for workspace: context deadline exceeded
* [preview] retry installing trust-manager
And use trust-manager from the packer image
* [test] clarify USER_TOKEN value for preview environments
* Cleanup
* [preview] remove commented out yaml related to PodSecurityPolicy
* [preview] force gce, no harvester
* [preview] default to using spot for GCE VMs
* [preview] bump sizes up now that we're using spot
...and actually use spot by changing the default for gce_use_spot
We'll address the disk pressure issues (which cause pods to evict on install, but eventually succeed) in a separate PR.
