Meego Wiki
Views

User:Tgalvin/O.8 system tests

From MeeGo wiki
Jump to: navigation, search

Contents

0.8 System Tests

Status

The status of 0.8 wrt ots.system_tests (Updated 2010-01-06)

Name Status
test_testrun_with_test_definition_tests See "Image Problems" below
test_host_based_testrun_with_test_definition_test See "Image Problems" below
test_hw_and_host_based_testrun_with_test_definition_tests
test_bad_image_url Works after minor fixes*
test_timeout Works after minor fixes*
test_non_existing_devicegroup Works after minor fixes*
test_non_existing_sw_product Works after minor fixes*
test_bad_testpackage_names Works after minor fixes*
test_no_image_url Works after minor fixes*
test_bad_distribution_model Distribution model needs fixing first
test_perpackage_distribution_no_packages Distribution model needs fixing first.
test_multiple_devicegroups Not done yet
test_one_devicegroup_multiple_devicenames Not done yet
test_one_devicegroup_one_devicename_multiple_device_ids Not done yet
  • Return values need clarifying (see below)

Current Issues

Return Values

I'm ashamed to say that I don't understand the return values CITA giving the overall status of the Testrun.

  • There are three possible ret vals "PASS", "FAIL" and "ERROR"
  • If there is an Error in the run (no image for example), The ots.system_test expects a "FAIL"

This behaviour of the Python unittest makes sense to me and it seems that in some respects the logic in CITA is the same e.g.the example above. But the current "ERROR" behaviour is unclear.

Passes Errors Success
False False False
True False False
False True False
True True True

Image Problems

I'm having problems running the tests with the image provided; ots_system_test_image.tar.gz. Being able to identify and communicate the images we are working with would seem necessary - do they have a hash?

test_testrun_with_test_definition_tests

On the image provided I get this error when running the generated command through the conductor direct:

 File "/home/tom/projects/ots/ots.worker/ots/worker/command.py", line 118, in execute_with_retries
   raise FailedAfterRetries

FailedAfterRetries 2011-01-05 09:09:09,015 INFO Testing in Hardware ended with error

test_host_based_testrun_with_test_definition_tests

Running this hangs...

0:06:00.086145 	2011-01-05 19:23:14.224707 	tom-desktop 	executor 	WARNING 	Retrying (1) to fetch file after (5) seconds sleep...
0:06:05.772793 	2011-01-05 19:23:19.911355 	tom-desktop 	executor 	WARNING 	Failed to fetch file /var/log/syslog (command scp root@192.168.2.15:/var  /log/syslog /home/tom/conductor/a37fe3b418ef11e0b2dd0025648621f1/Host_Hardware/post returned 1)
0:06:05.946484 	2011-01-05 19:23:20.085046 	tom-desktop 	executor 	WARNING 	Retrying (2) to fetch file after (5) seconds sleep...
0:06:11.633024 	2011-01-05 19:23:25.771586 	tom-desktop 	executor 	WARNING 	Failed to fetch file /var/log/syslog (command scp root@192.168.2.15:/var/log/syslog /home/tom/conductor/a37fe3b418ef11e0b2dd0025648621f1/Host_Hardware/post returned 1)
0:06:11.782244 	2011-01-05 19:23:25.920806 	tom-desktop 	executor 	INFO 	Storing testrun log file /var/log/testrun.log

A similar thing happened using the generated command and running on the Conductor:-

2011-01-05 18:07:04,394 WARNING Retrying (1) to fetch file after (5) seconds sleep...
2011-01-05 18:07:19,550 WARNING Failed to fetch file /var/log/syslog (command scp root@192.168.2.15:/var/log/syslog /home/tom/conductor/a37fe3b418ef11e0b2dd0025648621f1/Host_Hardware/post timed out)
2011-01-05 18:07:19,702 WARNING Retrying (2) to fetch file after (5) seconds sleep...
2011-01-05 18:07:34,855 WARNING Failed to fetch file /var/log/syslog (command scp root@192.168.2.15:/var/log/syslog /home/tom/conductor/a37fe3b418ef11e0b2dd0025648621f1/Host_Hardware/post timed out)
2011-01-05 18:07:35,020 INFO Storing testrun log file /var/log/testrun.log

Defect Driven Improvement

Problem Reporter Exposed by Description Short Term Resolution Long Term Recommendation New unittest
File recognition TG soak_test.py Results XML recognition by magic name "tatam_xml_testrunner" Patch with 0.1 implementation Formalised file transfer mechanism (e.g Mime Types and xsi instance in XML)
Response Queue TG soak_test.py Worker tries to post to response queue that isn't there Non-assertive posting for logging. Merged back in 0.1 Tighten protocol for Queue shut down. Robust behaviour for orphaned workers test_no_response_queue
Conductor Commands TV ots.system_tests Regression on command construction Code fixes Conductor will have an API in 0.9 test_conductor_commands_regression
Publisher Bootstrapping TV ots.system_tests http logs not seen when Value Errors on input parameters before Publisher boot Sandboxing and forced adaptation of Publisher parameters Is this a reasonable requirement. Can we refine the Publisher API to avoid this kludge test_hub.TestHubFailSafePublishing
Orphaned Messages E-P soak_tests.py Timed out messages picked up by new workers leading to Orphaned worker Not addressed! (Not arising from changes and no reported cases in Production) e.g. System synchronised clocks and "best before" stamp on Task Message
Missing Logs tv ots.system tests Essential log messages missing Reinstated N/A django_ots.TestView
Result return Value tv ots.system tests WIP
Personal tools