Bad Block Instructions - mdadm

[Bryan Larsen]

The Bad Block Howto should probably contain an mdadm section. mdadm will automatically repair bad blocks, so the section would be pretty short. Here's a first stab:

Linux md/raid will automatically repair a bad block on read by rewriting it. The best way to do this is by performing a scrub on the array:

# echo check > /sys/block/md0/md/sync_action

Follow the progress of the scrub by checking /proc/mdstat:

# cat /proc/mdstat

[Alex Samorukov]

I think we should wikify this page and open it to registred users to allow edit. There are a lot of things outdated or not mentioned in this doc, included ZFS, mdraid, SSD specific hints, etc.

[Ch.Ris]

The Bad Block Howto is also a good place to mention that the HDD erc timeouts must be set shorter than the controller timeout, to avoid a controller reset (causing loss of all unwritten data and kicking the disk out of the array, possibly making the array unusable) once a bad block is actually found (instead of getting the bad block remapped).

[Alex Samorukov]

I am agree with suggested changes. Bad block howto is now a wiki page with edit rights for the registered users. Please make your changes on https://www.smartmontools.org/wiki/BadBlockHowto directly. Thank you )