ostf tests rely on controller:/root/openrc but dont validate

[Andrew Woodward]

The ostf health tests rely on controller:/root/openrc, however the file can be changed by the sysadmin and could be wrong causing ostf to fail.

1) Since the credentials are known to the DB during deployment, an test should be performed to check if the openrc file and DB settings are not the same, this should result in a WARNING (not failure) that there may be a credential problem.

2) In the case of multiple controllers, a new test should be created where all controllers openrc files should be compared, they must contain the same values or result in a failure

3) A new test should be created to validate the credentials in use and fail with a reasonable description of the problem.

[Tatyanka (tatyana-leontovich)]

Hi, Andrew
Thank you for this point, but OSTf works in other way.

The ostf health tests rely on controller:/root/openrc - OSTF does not know about this file, It uses credentials from nailgun api.

But you point a right things as a what will be happened if admin change this file and does we should provide such possibility somewhere else (we provide it before deployment)

2) In the case of multiple controllers, a new test should be created where all controllers openrc files should be compared, they must contain the same values or result in a failure - can extend our HA non destructive tests with this scenario

A new test should be created to validate the credentials in use and fail with a reasonable description of the problem. - For credentials that nailgun provide - we already do such things. As about openrc - Lets discuss this on irc with the all team to decide what is the correct behaviour