Arvados

qr1hi-d1hrv-07ek396rvbutqjb is one instance that shows the problem, qr1hi-d1hrv-kd3498krej4o6ln is another.

I can see the normal pipeline instance page now, but I cannot cancel any jobs. I can pause though.

Edit: Was on the wrong account

Refreshing a running pipeline instance page reproduces this /sometimes/

• This issue is happening for pipelines in Running state. It is due to a recent update to handle start_at timestamp in apps/workbench/app/views/pipeline_instances/_running_component.html.erb

apps/workbench/app/views/pipeline_instances/_running_component.html.erb

           <% if current_job[:state] == "Queued" %>
             <%# column offset 5 %>
             <div class="col-md-6">
-              <% queuetime = Time.now - Time.iso8601(current_job[:created_at]) %>
+              <% queuetime = Time.now - current_job[:created_at] %>
               Queued for <%= render_runtime(queuetime, true) %>.
               <% begin %>

• Reverted that particular update as shown above

• Added an explicit test to verify the time display for a pipeline in Running state

• All pipeline instances test passed. Manual verification passed.

Added a test in 4393148 that passes with a1d70a2 but not without it. Using Time.parse(current_job[:created_at].to_s) seems to address the current bug without reinstating the older bug that (I'm assuming) motivated a1d70a2.