Who is SMART and who is not?

[Segfault]

I have this backup computer alright. I fire it up once a week and run a backup script to collect everything I care about on my home LAN. It has a super old 6 GB SSD for OS and four spinning WD red drives in RAIDZ. One of spinners did not pass SMART test today. It errored out at sector 600088 and the test terminated itself without completing. (Why, WD, why can't it carry on?) But it did nothing to reallocate the sector(s). OK, I thought, if I read that sector then the drive should notice it is bad. No such luck. My read test revealed 6 bad sectors in a row. Rest of the drive was good. Hmh, what if I write these six sectors. I figured since they are bad anyway my writing can't make it worse, and RAID should recover if there was data. I wrote 8 sectors, one before and one after the bad ones. The result confirmed 6 sectors out of 8 bad. To my surprise now the test passes, but reallocated count is still zero.
Apparently the drive is smarter than I am because I do not understand what happened.
Maybe there is someone here even smarter than the drive (and certainly me) and can shed some light on this? I am curious. Yeah, I know, cat and curiosity ... but still?

 

[likeman]

Segfault

on a HDD and SSD the only way you can fix a bad sector is to write to it (raid hardware and software raid do this automatically as long as you have RAID1 or higher) if the write succeeds and ECC Read verifies the data is intact (and signal level is good) it won't relocate it if it fails it will remap the LBA sector to the relocated area (if the Sector fails to remap to spare area the RAID will add it to the bad block table)

 

[Segfault]

Thanks, your reply put my mind back on this issue. I was wondering why these sectors do not show up as reallocated (reallocated count is zero). Then I realized my writing to these sectors may kind of resurrected them, and so it is. I tried to read these sectors and there is no error(s) any more, also smartctl test passes without errors. What exactly happened at low level is not clear to me, and I think the problem may return.

 

[likeman]

Thanks, your reply put my mind back on this issue. I was wondering why these sectors do not show up as reallocated (reallocated count is zero). Then I realized my writing to these sectors may kind of resurrected them, and so it is. I tried to read these sectors and there is no error(s) any more, also smartctl test passes without errors. What exactly happened at low level is not clear to me, and I think the problem may return.

normally it show as as pending relocation (usually) a read cant trigger a relocate so your count will be zero

when you write to that sector it will first attempt to write to that sector and then verify it, if it fails or signal level is to weak it will relocate it (if it passes the read verify it wont relocate it)

somthing might of happened like a power loss while it was writing data witch caused pass problems or vibration can cause problems as well (why nas/enterprise HDD have Rv sensors to correct for vibrations)

 

[Segfault]

Well, in old times we had low level format which allowed to fix issues. It was nothing else than writing the whole drive, but it often restored bad sectors. My drive still has no reallocated nor pending sectors.