From: alpine@vodafonemail.de
To: u.alpine.devel@lists.alpinelinux.org
Message-ID: <a710e79709ac4c66b51b39af6846e505@vodafonemail.de>
Date: Sun, 07 Feb 2021 12:08:18 +0100
Subject: too early bus timeout resets / data curruption instead of completing block-fail retries
MIME-Version: 1.0
Content-type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable





Hello,

are there alpine developers that would like to ship a solution to this
problem with alpine?

I've found a lot of discussion [1] around redundant setups (md, zfs, bt=
rfs,
rolling backup disk syncs) but the default 30 second bus timout will al=
so
completely reset about all standard single drives, before they would=20
return proper single-block reading retry success or error=2E

The solution seems to just be a matter of shipping udev rules that make
sure the bus timeout (/sys/block/sdX/device/timeout) is longer than the
drive's retry timeout=2E

Looks like the latest proposed udev rules have been uploaded as attachm=
ent
here:

https://www=2Esmartmontools=2Eorg/ticket/658
"Many (long) HDD default timeouts cause data loss or corruption
(silent controller resets)"

---

[1] For example:
RAID, SMART timeout (STCERC) and drives=2E
https://forum=2Eopenmediavault=2Eorg/index=2Ephp?thread/25398-raid-smar=
t-timeout-stcerc-and-drives-how-to-set-them-correctly-in-omv/