Currently, all our jobs running the functional tests in the Heptapod images go through the data initialization sequence of an Heptapod instance: creation of service configurations, database schema etc. This takes about 5 minutes
docker-inside jobs add the installation of chromium-driver on top of that, so that heptapod-tests can run inside the image. This is necessary for the backups tests, for instance.
In this MR, we create two new images, built with Kaniko, in which all of this is already done:
bootstrapped-chromium. On freshly produced images (by the omnibus-heptapod pipeline), this won't change the latency. But it will save at least 25 minutes of CI time for other jobs to run, especially when our container image caching plan is in service.
Of course, if the base image has not changed, it is expected to save 30 minutes of total CI time and 5 minutes of latency.
We can perhaps in a follow up also reduce the extra startup waiting time (currently 3 minutes).