Make the tests in arvados/sdk/cli/test to pass. Or skip the ones that do not pass.

http://yehudakatz.com/2010/12/16/clarifying-the-roles-of-the-gemspec-and-gemfile/

The development of Node Manager went like this: * First, it immediately paired cloud and Arvados nodes when booting them, to avoid duplicate assignment. * This didn't prevent duplicate assignments in cases where the cloud node struggled to come up, so we added the assigned_at logic to cover those. * We added the separate "booted" state to nodes to better handle eventual consistency in the clouds. Taken together, the logic to immediately pair nodes on boot is not only unnecessary, but causes trouble. A ComputeNodeMonitorActor for a node that doesn't bootstrap correctly may make incorrect shutdown suggestions based on the data in the Arvados node. Taking out the immediate pairing logic will make it easier to identify when a node has failed to bootstrap, and shut it back down appropriately.

It only has to do this on the dispatch node, not the individual compute nodes, if that helps.

We don't want to support this because we want to keep the compute nodes mutable. It's too difficult to ensure that the compute nodes will have the setup necessary to run a specific SDK version.

Mostly, this implements the actual work part of #4027, where Crunch installs an SDK version specified by the job record, and uses it to run the Crunch script. I tested this in a local Docker installation, using a Crunch script that reported its environment, and saved that as its output. In order to do that, I had to fix the Gitolite setup in our Docker setup to make repositories usable (making sure the API server is always SSH-able, and making a couple minor bugfixes to our Gitolite setup script). The script reported results as expected running outside Docker, inside Docker with no SDK specified, and inside Docker with an SDK specified. That script: <pre><code class="python">#!/usr/bin/env python import os import subprocess import arvados from pprint import pprint this_task = arvados.current_task() out_coll = arvados.CollectionWriter() with out_coll.open('environment.txt') as out_file: pprint(arvados.__file__, out_file) pprint(os.environ, out_file) if 'CRUNCH_INSTALL' in os.environ: pipe = subprocess.Popen(['ls', '-lR', os.environ['CRUNCH_INSTALL']], stdout=subprocess.PIPE) for line in pipe.stdout: out_file.write(line) pipe.stdout.close() this_task.set_output(out_coll.finish()) </code></pre> Please see the commit messages for additional information.

Another one liner!

The instructions for adding the "All users" class should include specifying "group_class":"role", otherwise the "all users" class doesn't show up in the "Groups" list.

At commit:3374f67bfe: Updated the help modal dialog to use the following links for API version and Workbench version: <pre> https://github.com/curoverse/arvados/tree/#{api_version}/services/api https://github.com/curoverse/arvados/tree/#{wb_version}/apps/workbench </pre> Refactored some code into a VersionHelper to simplify testing. Also updated the Rails.configuration.source_version settings for both API server and Workbench to strip trailing newlines (which were screwing up the URLs).

See story for details.

Just one line change.

Please see the git logs for detailed rationale.

LGTM, but please pull the branch before you merge; I made a few additional changes.

commit:874c563 Fixes: <pre> is_admin$ arv node list --select '["domain","job_uuid"]' Error: #<ActiveModel::MissingAttributeError: missing attribute: job_uuid> </pre>