I think the cable worked loose due to vibration of the hdd in an adjacent bay, and I'll use a locking cable if it happens again

I've purchased 14 SSDs, both 2.5" and M.2 form factors. I bought the first 30 months ago. Only one is in a 24/7 system. The rest get little usage. I've had to have a Mushkin 120G, a PNY 120G and an XPG (Adata) 128G, all 2.5", replaced under warranty. Life of each of the failures was significantly less than 12 months. A fourth, the replaced PNY, was in use about 10 months before hdparm -t performance dropped by about 90%. I got an RMA for it, but before returning, it I wiped it and repartitioned, and the hdparm -t performance returned to original. I told PNY about this, and it revoked the RMA.

1 members found this post helpful.

AIUI the system should run fstrim on an SSD device routinely. Failing that it will fragment/slow. You might check to see that the system that was running the PNY has that enabled.

On my fedora system that is (with systemd) fstrim.service.

Fstrim.timer runs weekly here. I used to have fstrim as a mount option instead, but the date long ago that that happened got lost when I wiped the PNY's partition tables and wrote new ones.