* retry functionality for lifestyle e2e test
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* attempt multiple cluster allocs
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* delete cluster before retry
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* fix
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* better output
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* retry allocate in all workflows
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* add a timer
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* add timers to ci
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
---------
Signed-off-by: David Fridrich <fridrich.david19@gmail.com>
* Remove usage of ttl.sh
Signed-off-by: Matej Vašek <mvasek@redhat.com>
* Fix detection of host resolution failure
Signed-off-by: Matej Vašek <mvasek@redhat.com>
---------
Signed-off-by: Matej Vašek <mvasek@redhat.com>
The current binaries and allocation scrips are very CI-centric, this PR
takes a few steps to make them more friendly to running in a dev
environment; mainly centered around not polluting global scope (keeping
filesystem mutations in a relative local directory).
- binaries (and kind kubeconfig) are installed in ./hack/bin
- adds /hack/bin to .gitignore
- binary paths can be overridden via env vars FUNC_TEST_{name}
- binary fallback is system-installed (when binaries.sh not run)
- Cleanup of the UX, including colors and some shared helpers via
library.sh
- Removes `yq` as the hack that required it appears to no longer be necessary.
- Adds some notes for Nix users
- Better error messaging when !linux
- Refactors the installers to pipe directly to tar rather than touch
disk.
- begin to standardize on hyphens in filenames
* test: use custom buildah image in tests
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* refactor: rename
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* test: use internal registry in on cluster tests
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* chore: more verbose output
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* fix: use fake ctx until fixing termination properly
Signed-off-by: Matej Vasek <mvasek@redhat.com>
---------
Signed-off-by: Matej Vasek <mvasek@redhat.com>
Use a different approach to pass current func image to the tests, this
approach does not require modification of task yaml nor access
to outer internet (ttl.sh).
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* chore: bump faas-js-runtime and cloudevents for node & TS
The Node.js and TypeScript templates have been bumped to the latest
versions of faas-js-runtime (2.2.2) and CloudEvents (7.0.1).
There is an issue with the `expectType` test for TypeScript CloudEvent
functions which I can't quite figure out, but the type test is not
critical and does not do anything to validate other than whether the
function signature is what's expected by faas-js-runtime, and in this
case, I think type exported type for faas-js-runtime may actually be the
issue here.
Signed-off-by: Lance Ball <lball@redhat.com>
* chore: pin all github actions to go 1.20.2
There was a change in how go parses URLs in versions higher than 1.20.2
(likely 1.20.6 and up) which breaks the Docker API's handling of daemon
URIs. This fix pins the Go version at a version known to work with the
existing Docker API and `func` implementation.
Signed-off-by: Lance Ball <lball@redhat.com>
---------
Signed-off-by: Lance Ball <lball@redhat.com>
* cleanup: style
Signed-off-by: Matej Vasek <mvasek@redhat.com>
* fix: use current func image for in cluster tests
Signed-off-by: Matej Vasek <mvasek@redhat.com>
---------
Signed-off-by: Matej Vasek <mvasek@redhat.com>
Luke Kingland