it is needed to have a possibility to always run a step in routine and also make it ignore non zero exit code.

In order to isolate default config, for example default config for specific openstack region, like 'network', 'floating' network, cloud name, flavour, as well as keyname, to allow user tweak according to their requirements

At the moment wasser only creates a single node cluster before running workflow.
It would be great if it could create multiple nodes, and allow user to run steps on any of it via labels.

investigate possible to safe more advanced logs to file simultaneously with non-verbose logs in console

It's a bit challenging to track down how and where the tests are running. Adding a quick-start type of document (and correcting URLs to ceph-ci-suse.de) would help others monitor jobs and lower the bar for contributions.

If no machines specified their is created only one server where routine is executed, each routine is supposed to have own server created. However some routines may have several server defined, so they can be created allocated in parallel or sequentially at very beginning of the routine execution.

by default we have a dictionary of routines, which are supposed to be executed in promiscuous way
workflows should allow to define order or parallel execution of routines