e2e: fix runtime reproducibility test and also check kata.contrast-node-installer-image

[miampf]

Follow up of #947. New successful run can be found here.

[katexochen]

@miampf could you separate the fix and the additional test into two distinct commits please (and rebase-merge later)?

[miampf]

@miampf could you separate the fix and the additional test into two distinct commits please (and rebase-merge later)?

It will take some time until merge anyways since the test is not failing as expected. When it comes to merging it I will keep this in mind tho :)

[miampf]

The test does fail if different values are detected (look here, I just randomly generated strings instead of using a hash), however, it does not fail when comparing the images digests (look here), reason being that they are equal. @katexochen, since reproducibility should be broken but the digests are the same, should we compare something other than the image digest? If yes, what do you think would be a good option?

[katexochen]

katexochen, since reproducibility should be broken but the digests are the same, should we compare something other than the image digest? If yes, what do you think would be a good option?

It's fine, the current reproducibility issues are really hard to reproduce, it would be to much work to detect that in a ci job. The test succeeds as this reproduces between the two machine, so there is no blocker to merge this from my perspective.