The general meaning of the SMART parameters is well documented, but a lot of noobs are misinformed. Not only are many raw parameterd vendor specific, but even ones that aren't are not always defined preciecly the same. Also different vendors include different attributes in their controllers.
I'd personally appreciate if you would divulge the disk make/model for each, but that's just my personal interest cropping up.
So first study the actual meaning of the raw attributes before jumping to conclusions.
Like #195 "Hardware ECC Recovered" - (Vendor specific raw value
.) The raw value has different structure for different vendors and is often not meaningful as a decimal number
Note that item #1 is vendor specific too!!
Google did a report and found that no
SMART parameter is a good indicator of impending failure. So I tend not to get my panties in a bunch just b/c the SMART parameters look a bit sour. Look at the "NOrmalized" and the Worst" and as long as these are well above the "Threshold" then the vendor finds the parameter OK.
So my reading is:
disk-01 - ~170 days uptime and ~175 power cycles, a large number of ECC correct and a 25 UNcorrected errors.
disk-03 - ~170 days uptime and ~175 power cycles, a large number of ECC correct and a 9 UNcorrected errors.
disk-02 & 04 - ~12 days uptime and time bignums for the vendor specific values - but no normalized or worst close to and thresholds
Well I'm sort of disturbed by drive 01 & 03, attribute #187. If you didn't have a mirror you'd have seen ~25 unrecoverable reads in ~6 months on -01. OTOH so long as these are in mirror they are pretty safe. You are quite unlikely to get coincident block errors on both copies of the mirror.
This is another opinion on attribute 187,
Personally I'd run both the short and long self-tests on both drives, and perform write-reads of the entire surface. If the vendor has a test tool (like seatools, or the WD equivalent) I'd run that too. That takes a long time for a 2TB drive. If the SMART error logs look bad - then replace.
The 'smartctl' command is your friend for the tests and logs, but I don't know enough to guide when they are behind a RAID controller.