Details
-
Type:
Bug
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: ts_middleware
-
Labels:None
-
Epic Link:
-
Sprint:TSSW Sprint - Sep 13 - Sep 27
-
Team:Telescope and Site
-
Urgent?:No
Description
There a few issues with the SAL-generated AuthList testing scripts
- The controller is started as a background process, but the script never terminates it, so it's left running as an orphaned process. The test script should terminate the controller before finishing.
- There is a line of text missing between the 7th and 8th test cases in the C++ test script. The "=======" separator needs to be added just before the "echo "Test with authList authorizedUsers=user@host,user2@other, nonAuthorizedCSCs=MTM1M3,MTM2,Test identity=MTM2"" line.
- The Script CSC does not support the "enable" command, so using the "enable" command as the test commander fails (for Script only). I suggest using one of the two generic commands that Script does support; maybe setLogLevel.
- The Java pathing isn't consistent. The salgenerator maven step accepts the 'version' parameter and uses the value in the ts_sal/test/maven/<CSC>-<XMLversion>_<SALversion><version> folder. The salgenerator java step also needs to support the 'version' parameter, so the path assignments are consistent.
Attachments
Activity
I figured out the reason why some of the tests are behaving differently for different CSCs. It's actually only behaving differently for the Test CSC, since that is used as one of the rotating authorized/non-authorized CSCs. For example, the sixth test case, "Test with authList authorizedUsers=user@host,user2@other, nonAuthorizedCSCs=Test identity=user2@other," assigns the Test CSC as a nonAuthorizedCSC. Obviously, this would behave differently for the Test CSC as it does for the other CSCs.
I think this was cleaned up last sprint
Dave Mills
added a comment - I think this was cleaned up last sprint The issues described in this ticket have all be addressed.
Also, the tests are not running consistently. I've seen different behavior between runs and CSCs, where sometimes the same tests "pass" with "Not permitted by authList" and with "completed ok." It would be nice to have some indication in the test script of the expected results. At least then I can investigate odd behaviors.