radhika chippada [Sun, 23 Mar 2014 21:20:37 +0000 (17:20 -0400)]
Raise exceptions on errors such as 'All users' group does not exist and a vm for the vm_uuid param does not exist. Add additional functional tests and update previous tests accordingly.
Brett Smith [Fri, 21 Mar 2014 19:00:15 +0000 (15:00 -0400)]
workbench: Add developer docs.
This adds a new README with information that's useful for people who
want to develop Workbench. Right now it's just information about how
to run and write tests. We can add to this as we go.
Some of this information was pulled from the installation guide. I
also deleted Rails' stock README while I was at it.
Peter Amstutz [Fri, 21 Mar 2014 17:02:28 +0000 (13:02 -0400)]
Merge branch 'master' of git.clinicalfuture.com:arvados
# Please enter a commit message to explain why this merge is necessary,
# especially if it merges an updated upstream into a topic branch.
#
# Lines starting with '#' will be ignored, and an empty message aborts
# the commit.
Peter Amstutz [Fri, 21 Mar 2014 16:57:04 +0000 (12:57 -0400)]
Merge branch 'master' of git.clinicalfuture.com:arvados
# Please enter a commit message to explain why this merge is necessary,
# especially if it merges an updated upstream into a topic branch.
#
# Lines starting with '#' will be ignored, and an empty message aborts
# the commit.
Brett Smith [Fri, 21 Mar 2014 14:26:16 +0000 (10:26 -0400)]
workbench: Improve API server integration launch.
The previous version sometimes tried (and failed) to find the server
pid before rails wrote it. This provides some limited retry, and
avoids spawning a subshell too.
Brett Smith [Fri, 21 Mar 2014 14:25:26 +0000 (10:25 -0400)]
workbench: Improve integration test environment.
Removing too much from the environment causes builds to fail in
Jenkins. Instead of using a whitelist of environment variables, use a
blacklist of variables that Bundle modifies.
Brett Smith [Thu, 20 Mar 2014 21:24:24 +0000 (17:24 -0400)]
workbench: Make integration test settings default.
We know that these settings should work out of the box, given the
settings in the API server's `rails` script. Making them the default
makes it easier to get the tests running.
Peter Amstutz [Thu, 20 Mar 2014 19:23:56 +0000 (15:23 -0400)]
- Made a note about intentionally not setting "success"
- Changed if c[:job][:success] == false to elsif
- Changed == nil to .nil?
- Removed whitespace
Brett Smith [Wed, 19 Mar 2014 16:10:51 +0000 (12:10 -0400)]
workbench: Add VM integration tests.
This is the "start test" that Ward suggested I implement to start off.
It's easy but not trivial: it requires JavaScript support and filling
out a form.
Brett Smith [Wed, 19 Mar 2014 15:44:49 +0000 (11:44 -0400)]
workbench: Add integration test helper module.
This provide a base test class that provides the Capybara DSL and a
couple of convenience methods. It also specifies a test runner class
that sets up the API server consistently, with the proper environment.
Peter Amstutz [Mon, 17 Mar 2014 17:49:06 +0000 (13:49 -0400)]
* Allow the system user to update job columns, even if the job is "locked"
* Once the job process completes, automatically mark "job.running = false" and set "job.finished_at" if not set
* Exclude jobs from the queue where the "success" field is already set
* Consider a pipeline to be failed if any of the component jobs are marked as failed
Tom Clegg [Mon, 17 Mar 2014 16:56:57 +0000 (12:56 -0400)]
Use --no-clear-tmp flag to decide whether to remove of old temporary
directories, rather than deciding based on whether script_version is a
commit or a local directory.
Brett Smith [Fri, 14 Mar 2014 17:34:29 +0000 (13:34 -0400)]
api: Improve compute_node_domain handling.
The default value was causing test failures on my machine because the
ERB evaluated to an empty string. And we want to avoid this causing
any DNS lookups during tests. Now we make the default false (in line
with our other compute_node_* settings), and suggest a change in
application.yml.example to demonstrate how it should be set.
I also fixed up the example about compute_node_nameservers while I was
at it.
Peter Amstutz [Thu, 13 Mar 2014 21:07:46 +0000 (17:07 -0400)]
* Added support for offset and limit query parameters (?limit=x&offset=y) to workbench index pages.
* Created _paging partial for rendering forward/backward/jump to page N
* Added paging to generic index page, index pages for collections, pipeline instances, pipeline templates, groups.