I think the reason was that we had most servers licenced (enterprise repo) but our two new servers we didnt have licenced yet.
We then licenced them a week ago but didnt reboot them. When we tried to replace osds it kept freezing looking at logs at some point in the creating and formating of the OSDs. I see one of us tried it with two different ssds.
When then rebooted those two nodes that were added after updating and it creation then worked fine.
I did ceph osd purge osd.23 and ceph osd purge osd.25 and it worked fine.