A couple of days ago system SSD in my desktop failed completely. A week or two before some symptoms appeared: the system occasionally booted as read-only, some errors appeared. But after a reboot all worked well. SMART didn't show anything at all.
This is my main working computer and it is 10-year-old (with some upgrades). I have proper backups set up to a HDD and also important data is sent to a cloud in addition to the HDD.
So I thought: "Okay, I am too busy now, the system is failing but working. Maybe it is the SSD, maybe the motherboard or power supply is dying. I have backups. I'll keep working for some time and if it dyes completely, I replace the SSD or the whole system".
After it finally dyed completely (it was very inconvenient at that time), I found out that it was the SSD, bought another one very fast, restored a backup and was happy enough.
But still I lost some portion of my work (not a big one but still) between the backups.
I am using btrfs for a long time. I thought that there must be a better way of doing it with using raid1. So I bought another SSD and made a raid.
The plan was in case an SSD fails to mount the raid with
degraded option, convert it to
single, order a disk and keep working. If the other SSD doesn't fail right away, it should be good enough.
But there is a gotcha. Imagine the same failure case. How do I know which disk is guilty?
The problem is that raid1 can be mounted as degraded only once, so it is hard to test it.
Does any have any ideas?