[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-bcache
Subject:    Re: Stale bcache (writethrough) leading to cache being disabled.
From:       Thorkild Stray <thorkild.stray () cxense ! com>
Date:       2014-04-11 1:10:30
Message-ID: CAJ0Z65y1fcszQXChAjbH72VCOYxsQzAhP6U=_X428mMZHsNBVA () mail ! gmail ! com
[Download RAW message or body]

Hi,

On Wed, Apr 9, 2014 at 6:05 AM, Thorkild Stray
<thorkild.stray@cxense.com> wrote:
> The specific kernel message when it disables itself is:
> 2014-03-25T23:55:03.525100+00:00 machine1 kernel: [6822460.984018] bcache: error on \
> 5acb2ff8-db07-4d13-a6ca-dbfdbf1d87eb: key too stale: 97, need_gc 97, disabling \
> caching

Upgrading to the newly released 3.14 kernel (no extra patches) and
resetting the caching device solved this (first reset, then upgrade).

To reset the cache in a cleanly unmounted setup without having to
reformat the backing device (requirement that you have no data left
to write back - in my case we use write-through so no issue there), do:

1) De-attach and unregister the device. Take note of the identifier.
(I also unmounted it to avoid any problems)
2) Stop bcache (echoing 1 into /sys/block/bcache0/../stop )
3) Run make-bcache --cset-uuid <same uuid as before> -C <device> to
re-init it. Do not touch the backing device.

At this point you should get "run_cache_set() invalidating existing
data" in the dmesg-output. It'll auto-register the new caching-device,
but you still need to re-add the backing device and re-attach the bcache-set.

In my case, bcache had already politely disabled itself due to the
error, so I could skip many of the initial steps.

-- 
Thorkild
--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html


[prev in list] [next in list] [prev in thread] [next in thread]