Please help us help you and fill where relevant: 
Your LimeSurvey version: LimeSurvey Community Edition Version 6.12.0 
Own server or LimeSurvey hosting: Own server
Survey theme/template: Default one
==================
Hi,
I have a set of 20 question groups and I want 2 questios groups to always come up (as my control) and 8 to be randomly selected from what is available.
So for example: User 1 sees question group 1, then 2 and then 12-20. And then User 2 might see  group 1, then 2 and then 3-10.
Is such a thing possible? 

Thanks a lot in advance for any help!
 

If the order of these two randomly chosen question groups does not matter, assign all question groups to the same randomization group.

Then set the relevance/Expression script  of all these question groups to 0, except for the first two ones, which you leave at 1.
Now the question groups in that randomization group will randomly switch places, but only two will be shown.

So even I am trying to do kind of same thing. I have 30 question groups and from that I want randomly any 10 question groups to be shown. Any random 10 from the 30 groups every time. 
How do I do this? I tried the way you suggested but then only first 10 are always shown and not the others

How do I do this? I tried the way you suggested but then only first 10 are always shown and not the others

Then you didn't do it right. Here again in simple steps:

1. Create your question groups
2. Asign them all to the same Randomization Group (you basically give the same name in "Randomization Group" for all 30 groups).
3. Hide all but the first 10 groups, as described above

What will Limesurvey do?

It will randomize the different question groups in the same "randomization group" among each other. So group 10 might be on position 24, group 30 might be on position 5, etc.

But the respondent will only see those questions groups that fall randomly into the position 1-10, because 11 to 30 are hidden.

If in your case it always shows only question 1-10, you most probably skipped the part with asigning all questions groups to the same randomization group.