How toast is my SSD, and why?

Solution 1:

The SMART status shows a lot of old or dying indicators, but nothing particularly screams " this killed it!".

Your log shows a power on lifetime of 995 days and 10 hours, suggesting that you leave your machine on permanently which isn't a bad thing of itself, it just means that the drive has seen a lot of hours of small writes as the operating system does book-keeping and general use.

To me it looks like the SSD is just old and worn out. The Perc_Rated_Life_Used is surprisingly low, as is Erase_Fail_Count

The worrying thing for me is your "regular" hitting of 95%+ full, which will reduce the pool of empty blocks available for the wear levelling algorithm to do its job. You will effectively end up stressing a small amount of blocks harder during the times when you are starved of space resulting in a small cluster of blocks with a massive level of writes, while the average across the drive is pretty low. By doing it repeatedly the wear leveller will probably choose the "best" (least written) blocks to write to first, but as you get to 100% full you are left with the "worst" blocks. Combine that with general programs and the operating system running its tasks means that you are going to be wearing out the worst blocks a lot faster. It's a perfect way to stress the worst parts of the drive and send it to an early grave.

You effectively force the key filesystem and SSD bookkeeping functions into the worst cells as they are likely to be written regularly to the drive, particularly when the SSD is mostly full, and sooner or later something bad is going to happen. If you've run out of reallocatable blocks and a key structure cannot be moved then the drive could deadlock itself.

This is why people say that you should always try to keep some anecdotal amount of space free on your drive, because the less space you have free the harder you are working the area that is free.

It's possible that old age and heavy writes to small groups of blocks have worn out parts of the drive.

Chances are copying what you need to a new drive will be fine, hardware failures like this tend not to be contagious.