very high IO utilization after upgrade from mongo 3.6 to 4.0.18

We upgraded our sharded cluster from mongo 3.6 to 4.0.18 on Linux Stretch. Since then IO utilization is almost always at 100%Unable to render embedded object: File ( Almost all monitored metrics look worse than before the upgrade) not found. I'll attach a screenshot where you can see that IO utilization went up to almost constant 100% at around 0:00 o'clock when we upgraded the shard to mongodb 4.0.18 

We did not yet set FCV to 4.0 to allow our deployment to run without enabling these features for a burn-in period to ensure the likelihood of downgrade is minimal - as suggested in mongodb's documentation.

Any help is greatly appreciated because this issue becomes critical!