GitLab

Update Kubernetes pull policy documentation

See merge request gitlab-org/gitlab-runner!2860

Document operator properties, custom installation and permissions

Closes gitlab-org/gl-openshift/gitlab-runner-operator#16, gitlab-org/gl-openshift/gitlab-runner-operator#20, gitlab-org/gl-openshift/gitlab-runner-operator#6, and gitlab-org/gl-openshift/gitlab-runner-operator#29

See merge request gitlab-org/gitlab-runner!2847

Allow user to specify multiple pull policies for Kubernetes executor

Closes #27298

See merge request gitlab-org/gitlab-runner!2807

ci: wrap GOCACHE with double quotes

See merge request gitlab-org/gitlab-runner!2859

Fix job duration value when in log format is json

See merge request gitlab-org/gitlab-runner!2787

We are seeing failures like
https://gitlab.com/gitlab-org/gitlab-runner/-/jobs/1159137557 and
according to
https://stackoverflow.com/questions/2743673/mkdir-error-in-bash-script
this is because we aren't wrapping variable in quotes.

Fix Kubernetes attach strategy for non-root environments

Closes #27107 and #26540

See merge request gitlab-org/gitlab-runner!2749

Fix ubi image tag

remove ubi image

remove test with int overflow

Closes #25765

See merge request gitlab-org/gitlab-runner!2597

Removed reference to master

See merge request gitlab-org/gitlab-runner!2855

Related to: https://gitlab.com/gitlab-org/gitlab-runner/-/issues/27732

Roughly alphabetised Docker Container parms - docs

See merge request gitlab-org/gitlab-runner!2851

Make clear when FF runner configuration syntax was introduced

See merge request gitlab-org/gitlab-runner!2852

Use inspect.GID() to collect GID value in Docker executor

Closes #27600

See merge request gitlab-org/gitlab-runner!2769

Report CI test build failures

Closes #27680

See merge request gitlab-org/gitlab-runner!2829

Fix pipeline configuration for docs branches

See merge request gitlab-org/gitlab-runner!2853

With https://gitlab.com/gitlab-org/gitlab-runner/-/merge_requests/2845
we've introduced a bridge job trigerring our integration tests in a
dedicated project. This however didn't include the case when docs branch
are pushed and the needed binaries job is not created at all.

This commit fixes this situation.

Signed-off-by: Raimund Hook <raimund.hook@exfo.com>

Compile gitlab-runner binary for shell integration tests

See merge request gitlab-org/gitlab-runner!2820

Add bridge job to runner-incept

Closes gitlab-org/ci-cd/tests/runner-incept#2

See merge request gitlab-org/gitlab-runner!2845

Documentation Update/docs/security/index.md

See merge request gitlab-org/gitlab-runner!2849

Updated docs reference to MinIO

See merge request gitlab-org/gitlab-runner!2850

The doc reference to MinIO (under
docs/configuration/speed_up_job_execution.md) did not specify admin
credentials forcing MinIO to use insecure defaults.

The reference further down in the same doc directed the user to `grep`
the keys from a json file which is now encrypted by default.
Signed-off-by: Raimund Hook <raimund.hook@exfo.com>

Adding git depth for submodules

See merge request gitlab-org/gitlab-runner!2107

Remove docker-machine provision on creation failure

Closes #27613

See merge request gitlab-org/gitlab-runner!2805

Update warning message URL for DockerHub

See merge request gitlab-org/gitlab-runner!2844

In
https://gitlab.com/gitlab-org/gitlab-runner/-/issues/27613#note_522938477
we see machines failing to create on the cloud provider, but it fails
for rate limit reasons or anything else. This leads to a split-brain issue where we have state of a machine "created" in `docker-machine`
state but the machine is nonexistent/inaccessible to the cloud provider.

If the creation fails `docker-machine provision` runs on a machine that
doesn't exist which wastes API requests, and time trying to wait for SSH
to be available on a machine. The `docker-machine provision` tries to
re-provision an existing machine, but if creation fails that means that
the machine doesn't exist.

Instead of retrying to create/provision the machine with the same details
fail fast, remove the machine and try to create a new one with details.
This should result in much quicker recovery when we reach the API rate
limits on the cloud provider and also job queues because the machine
would be available sooner rather than waiting for `docker-machine
provision` to be available.

To enable this fail fast functionality you have to enable
`FF_SKIP_DOCKER_MACHINE_PROVISION_ON_CREATION_FAILURE` feature flag
inside of the `config.toml`

reference https://gitlab.com/gitlab-org/gitlab-runner/-/issues/27613