1 |
On Tue, May 19 2015, Rich Freeman wrote: |
2 |
|
3 |
> On Tue, May 19, 2015 at 11:22 AM, <gottlieb@×××.edu> wrote: |
4 |
>> On Tue, May 19 2015, Peter Humphrey wrote: |
5 |
>> |
6 |
>>> On Tuesday 19 May 2015 10:53:26 Rich Freeman wrote: |
7 |
>>>> On Tue, May 19, 2015 at 10:44 AM, Peter Humphrey <peter@××××××××××××.uk> |
8 |
>>> wrote: |
9 |
>>>> > Incidentally, what's the received wisdom on frequency of file-system |
10 |
>>>> > trimming on SSDs these days? I've seen values quoted between twice a day |
11 |
>>>> > and once a week. And how does trimming affect btrfs? |
12 |
>> |
13 |
>> I included "discard" in fstab for my ssd filesystems, presumably |
14 |
>> following some installation guide. For example I have |
15 |
>> |
16 |
>> /dev/sda5 / ext4 noatime,discard 0 1 |
17 |
>> /dev/vg/local /local ext4 noatime,discard 0 2 |
18 |
>> |
19 |
>> Is it preferred to instead issue explicit trim's via cron? |
20 |
>> |
21 |
> |
22 |
> It depends. |
23 |
> |
24 |
> In theory giving your drive useful information about allocation now is |
25 |
> better than giving it the information later. The drive can make use |
26 |
> of that information to improve performance. |
27 |
> |
28 |
> In practice some drives have brain-dead firmware and they'll do stupid |
29 |
> things with that information. If you trim part of an erase block, the |
30 |
> drive should just file that info away and make use of that information |
31 |
> when it can. However, some drives will immediately copy/erase the |
32 |
> rest of the block at that moment, which creates an unnecessary erase |
33 |
> cycle and creates IO load at a moment that the drive is already busy. |
34 |
> |
35 |
> So, if your drive isn't brain-dead discard is better. If your drive |
36 |
> is brain-dead fstrim is almost as good if the drive isn't too full. |
37 |
> I've yet to test discard and see how well it works. |
38 |
|
39 |
Understood. Thank you. |
40 |
allan |