Kernel/driver parameter to limit (S)ATA disk bad sector read retries?

Nyyr · 11-22-2012, 05:03 AM

Hello!

I have RAID1 mdadm setup (2 SATA disks) on Ubuntu 11.10 and one of the disks during boot causes system to freeze trying to read from the disk. After half an hour of waiting I had to reset the PC. When I change BIOS settings to boot from the other disk of the RAID1, it ends the same way.
(I then managed to run SeaTools and repair the disk (by reallocating unreadable sectors I guess) and now it is OK.)
So is there a way to tell kernel/driver to report the disk as failed after the first unsuccessful attempt to read the bad sector?

smallpond · 11-25-2012, 11:06 AM

You sure that's what's going on? No BIOS knows how to boot from Linux software RAID. That means you need to be booting to RAID-enabled GRUB, a copy of which has been installed on each disk in your RAID 1. GRUB will have to assemble and boot from your RAID. There are many variables involved:

Are you using GRUB or GRUB-2?
Are you using DOS partitions or GPT?

Here's a typical procedure for non-GPT boot:
http://www.texsoft.it/index.php?c=ha...d-squeeze&l=it

Nyyr · 11-27-2012, 10:08 AM

Yes, I´m sure what´s going on. It doesn´t matter what version of grub I´m using and partition type it not involved at all!!!
I suggest that you read CAREFULLY my post before replying. I said it freezes DURING boot and NOT that it doesn´t boot at all!!!

Even when not using RAID at all, the system should report an error reading disk and not stay stuck trying to read a bad sector for a half of an hour !!!!