mrsas(4) handling for large misaligned I/Os

Description

In TrueNas Web GUI, I tried to replace defective HDD on a ZFS pool. When I launch replacement operation, an error appear :

middlewared.service_exception.CallError: [EFAULT] Failed to wipe disk da5: [EFAULT] Command dd if=/dev/zero of=/dev/da5 bs=1M count=32 failed (code 1):
dd: /dev/da5: Input/output error

This error appear with two new Hard Disk Western Digital WD4003FFBX.

After debug this issue, I have discover the problem. When I launch DD command with a Block Size > 256K, error appear. But when I launch DD command on GNU with BS 1M, this error don't appear.

This DD error is only on TrueNas, FreeBSD system.

To replace my disk I must launch the replacement on CLI, but GUI is not fonctionnal.

I hope this review of a bug will help you to improve your system.

Problem/Justification

None

Impact

None

Activity

Show:

Alexander Motin July 1, 2021 at 7:47 PM

I've fixed the bug for next 12.0-U5 update.  If you are willing to test the change earlier, I can provide you a custom built kernel.

Alexander Motin July 1, 2021 at 6:05 PM
Edited

It is not a problem of the failed disk, but a problem of mrsas(4) RAID card driver you are using, improperly reporting its capabilities.  I see such errors in your logs:

Berlemont June 30, 2021 at 6:20 AM

I send you the Debug file. I hope this will help you.

Thanks.

iX-Tony June 29, 2021 at 6:57 PM

Thank you for the report . Is the GUI accessible? If so, can you please attach a debug file to this ticket? To generate a debug file on TrueNAS CORE, log in to the TrueNAS web interface, go to System > Advanced, then click SAVE DEBUG and wait for the file to download to your local system.

Complete

Details

Assignee

Reporter

Labels

Impact

High

Time remaining

0m

Components

Fix versions

Affects versions

Priority

Katalon Platform

Created June 29, 2021 at 3:23 PM
Updated July 6, 2022 at 8:58 PM
Resolved July 1, 2021 at 7:47 PM