* Input requirement for persistent configuration items in bcache
@ 2020-06-22 8:01 Coly Li
0 siblings, 0 replies; 4+ messages in thread
From: Coly Li @ 2020-06-22 8:01 UTC (permalink / raw)
To: linux-bcache@vger.kernel.org
Hi folks,
Recently I start to look at the bcache on-disk superblock extension. I
still remember years ago people discussed adding some persistent
configuration itesm on disk, to avoid setting them via sysfs interface
every time after reboot.
While I am thinking about this, I do appreciate your help about the
input on which configuration item should be stored in the on-disk
superblock.
Although I am probably not able to make every required input being
persistent in the on-disk superblock, I believe your help may help me to
decide a reasonable option set.
Thank you all in advance.
Coly Li
^ permalink raw reply [flat|nested] 4+ messages in thread
* Input requirement for persistent configuration items in bcache
@ 2020-07-05 4:45 Coly Li
2020-07-05 9:12 ` Matthias Ferdinand
0 siblings, 1 reply; 4+ messages in thread
From: Coly Li @ 2020-07-05 4:45 UTC (permalink / raw)
To: linux-bcache@vger.kernel.org
Hi folks,
Recently I start to look at the bcache on-disk superblock extension. I
still remember years ago people discussed adding some persistent
configuration itesm on disk, to avoid setting them via sysfs interface
every time after reboot.
While I am thinking about this, I do appreciate your help about the
input on which configuration item should be stored in the on-disk
superblock.
Although I am probably not able to make every required input being
persistent in the on-disk superblock, I believe your help may help me to
decide a reasonable option set.
Thank you all in advance.
Coly Li
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Input requirement for persistent configuration items in bcache
2020-07-05 4:45 Input requirement for persistent configuration items in bcache Coly Li
@ 2020-07-05 9:12 ` Matthias Ferdinand
2020-07-05 10:23 ` Coly Li
0 siblings, 1 reply; 4+ messages in thread
From: Matthias Ferdinand @ 2020-07-05 9:12 UTC (permalink / raw)
To: Coly Li; +Cc: linux-bcache@vger.kernel.org
On Sun, Jul 05, 2020 at 12:45:41PM +0800, Coly Li wrote:
> While I am thinking about this, I do appreciate your help about the
> input on which configuration item should be stored in the on-disk
> superblock.
Hi,
my use case is rather limited, all my bcache devices run in writeback
mode, and the only parameters I keep playing with are writeback_percent
and sequential_cutoff.
My short list of favourites:
- cache_mode
- writeback_percent
- sequential_cutoff
But I vaguely remember that I used rebooting as last resort after too
much fiddling with tunables; persisting them in the superblock means
losing this reset-to-defaults method. Would be nice to have some other
way of resetting.
An alternative to modifying the superblock might be something
more out-of-band like iptables does with iptables-save and
iptables-restore: convert current settings to text, e.g. on shutdown or
at regular intervals, and restore settings on boot from this text file.
Regards
Matthias
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: Input requirement for persistent configuration items in bcache
2020-07-05 9:12 ` Matthias Ferdinand
@ 2020-07-05 10:23 ` Coly Li
0 siblings, 0 replies; 4+ messages in thread
From: Coly Li @ 2020-07-05 10:23 UTC (permalink / raw)
To: Matthias Ferdinand; +Cc: linux-bcache@vger.kernel.org
On 2020/7/5 17:12, Matthias Ferdinand wrote:
> On Sun, Jul 05, 2020 at 12:45:41PM +0800, Coly Li wrote:
>> While I am thinking about this, I do appreciate your help about the
>> input on which configuration item should be stored in the on-disk
>> superblock.
>
> Hi,
Hi Matthias,
Thank you for the input.
>
> my use case is rather limited, all my bcache devices run in writeback
> mode, and the only parameters I keep playing with are writeback_percent
> and sequential_cutoff.
>
> My short list of favourites:
>
> - cache_mode
> - writeback_percent
> - sequential_cutoff
>
Yes I plan to define a very limited minimum configuration set. And
rested should be configured via user space tool.
> But I vaguely remember that I used rebooting as last resort after too
> much fiddling with tunables; persisting them in the superblock means
> losing this reset-to-defaults method. Would be nice to have some other
> way of resetting.
>
> An alternative to modifying the superblock might be something
> more out-of-band like iptables does with iptables-save and
> iptables-restore: convert current settings to text, e.g. on shutdown or
> at regular intervals, and restore settings on boot from this text file.
Yeah, I think in the same way as yours.
Coly Li
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-07-05 10:23 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-07-05 4:45 Input requirement for persistent configuration items in bcache Coly Li
2020-07-05 9:12 ` Matthias Ferdinand
2020-07-05 10:23 ` Coly Li
-- strict thread matches above, loose matches on Subject: below --
2020-06-22 8:01 Coly Li
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.