Suddenly very slow yesterday / Everything fine today

I use 4x 4TB WD Reds (WD40EFRX) in a software RAID (mdadm). Yesterday, the performance of the RAID dropped drastically. For testing, I copied a 60GB file from the RAID to the system SSD and it did not get over 25MB/s. iotop showed that the speed of the copy process was 0-25 MB/s.
I checked the SMART states of all 4 disks and started a short test. The test of 3 of 4 disks were stuck at 90% so that I assumed a defect.
Today, I wanted to investigate it in more detail but everything is fine … The same 60GB file was copied at ~300MB/s without any drops.
All SMART short tests finished without any errors.
The drives are 3 years old but were barely used (Power on hours ~200-250)
The number of Reported Uncorrectable Errors are 0 for all disks.

All SMART states are fine except:
“ID”,“Name”,“Failed”,“Norm-ed value”,“Worst”,“Threshold”,“Raw value”,“Type”,“Flags”
1,“Raw Read Error Rate”,“never”,“200”,“200”,“51”,“0”,“pre-failure”,“POSR-K”
3,“Spin-Up Time”,“never”,“163”,“163”,“21”,“6.816”,“pre-failure”,“POS–K”
5,“Reallocated Sector Count”,“never”,“200”,“200”,“140”,“0”,“pre-failure”,“PO–CK”
The values are exaclty the same for all disks. I read that these values are not very meaningful…

Are there any other SMART values that give me a clue what might have gone wrong yesterday?
Or do you have any idea what else could cause the problem or how to reproduce it?

All drives are connected to a ASROCK B460M board. The OS is Ubuntu 20.04. Furthermore, I have 3x optical BD-ROMs which are connected by an external PCI / SATA Controller