wiki:access/AccessDevQAChecklist

QA Checklist for Accessdev


Introduction

This wikipage sets out a QA checklist for testing the NCI virtual machine accessdev.nci.org.au. The purpose of the checklist is provide system administrators of accessdev a means of determing if the system is running correctly or not.

List of Components for Checking

The following components will be used to verify accessdev.nci.org.au or accessdev-test.nci.org.au is running correctly.

Id Component Comments
S1 Rose
S2 Cylc
S3 FCM
S4 UMUI
A1 Job submission
A2 Files systems mounted (production only)

Checklist for Job submission (A1)

Id Component Expected Output Pass Fail Comments
A1.1 Verify job submission works
$ remote-job-submission
$ ssh raijin.nci.org.au qstat
# From Raijin
$ ssh accessdev cylc

Checklist for File system mounts (A2)

Check filesystems: /home /scratch /g/data1 (not critical at the moment)

Checklist for Rose (S1)

Prerequisites

  • Job submission works correctly
  • ssh keys are set correctly

Checklist regarding Rosa and Rosie

Id Component Expected Output Pass Fail Comments
S1.1 Rosa Server is running
$ service rosa status
rosa (pid 6452) is running...
S1.2 Rose repositories are accessible
$ rosie ls
Printout of local suites. May not print anything if no local suites are checkout.
S1.3 Rosie go Interface works
$ rosie go
Rosie go interface should appear. Currently, there a several gtk warning that are printed out. This might be due to gtk2 2.20 being used, rather than gtk2 2.18

Checklist regarding running a suite under Rose

Id Component Expected Output Pass Fail Comments
S1.5 Run Testing suite
$ rosie checkout test-aa000
$ rose suite-run -C ~/roses/test-aa000
Cylc window should appear with tasks being submitted and executed Need a basic suite that can check to see if Rose is working

Checklist for Cylc (S2)

Prerequisites

  • Job submission works correctly
  • File cylc-test/suite.rc exists with contents
    title = "Remote job submission test"
    [scheduling]
        [[dependencies]]
            graph = "A => B"
    [runtime]
        [[root]]
            initial scripting = """
                module use /g/data/access/modules
                module load cylc
                """
            [[[job submission]]]
                method = pbs
            [[[directives]]]
                -l = "walltime=1:00,ncpus=1,vmem=500mb"
            [[[remote]]]
                host = raijin.nci.org.au
    
  • Register Job
Id Component Expected Output Pass Fail Comments
Register Job
$ cylc register test cylc-test
REGISTER test: /home/562/saw562/cylc-test If 'test' is already registered remove it with
$ cylc unregister test
  • Run under CLI
Id Component Expected Output Pass Fail Comments
Run Job
$ cylc run test
Summary screen with name, PID, Port and log directory Job A.1 should be in Raijin's run queue, once that has ended B.1 will be submitted
List Jobs Running
$ cylc run scan
test saw562 accessdev.nci.org.au 7767
  • Run under GUI
Id Component Expected Output Pass Fail Comments
Start GUI
$ gcylc test
GUI Screen appears
Run Job
Press 'play' button on toolbar, press Start button
GUI updates to show 'submitting' followed by 'submitted' status on job A
Wait for completion GUI updates to show 'finished' for job A then updates as job B is submitted and run

Checklist for FCM (S3)

Id Component Expected Output Pass Fail Comments
Show UM trunk
$ fcm ls - v fcm:um_tr
Show past UM trunk by version
$ fcm ls -v fcm:um_tr@vn7.3
Show past UM trunk by revision
$ fcm ls -v fcm:um_tr@1072
Show UM branches
$ fcm ls fcm:um_dev
Show JULES trunk
$ fcm ls fcm:jules_tr

Checklist for UMUI (S4)

Prerequisites:

  • Remote job submission (accessdev to raijin only)
  • FCM
Id Component Expected Output Pass Fail Comments
Run UMUI client:
$ umuix
UMUI database browser opens
accessdev umui server up:
Click on "*accessdev" tab of umui
Click on "All" tab to see other people's jobs
Jobs are visible
Connect to accesscollab umui server:
Click on "Collab" tab of umui
Click on "All" tab to see other people's jobs
Jobs are visible
Job opens:
Open any umui job
Jobs opens without any errors
Job runs:
Open a standard job (e.g. Collab:sabta)
Click process, then submit.
Jobs successfully extracts then submits job
Check for job on queue on raijin

Checklist for Certificates (A3)

Id Component Expected Output Pass Fail Comments
On accessdev:
$ curl -L http://accessdev.nci.org.au
$ curl -L http://accessdev-test.nci.org.au
Print out of HTML script Failure will produce a message like curl: (60) Peer certificate cannot be authenticated with known CA certificates
On raijin:
$ curl -L http://accessdev.nci.org.au
$ curl -L http://accessdev-test.nci.org.au
Print out of HTML script Failure will produce a message like curl: (60) Peer certificate cannot be authenticated with known CA certificates
Last modified 6 years ago Last modified on Oct 10, 2014 2:25:50 PM