Because of a general check for remaining strech machines across the cluster (T247045#8063348) I found that
ms-be2028 is still on stretch, while not a single other ms-be host is still on stretch.
Seems like this happened because this server had several hardware issues (T294001, T279245, T279644, T160349). It was probably out of service while the other hosts were all upgraded and then it was fixed later.
Now it still appears in reboot tickets and it's up but:
[ms-be2028:~] $ lsb_release -c
Codename: stretch
It should be upgraded (or decom'ed?).