1 |
Am 25.12.2012 19:26, schrieb Mark Knecht: |
2 |
> On Tue, Dec 25, 2012 at 9:18 AM, Michael Orlitzky <michael@××××××××.com> wrote: |
3 |
>> On 12/25/2012 12:07 PM, Mark Knecht wrote: |
4 |
>>> On Tue, Dec 25, 2012 at 8:33 AM, Pandu Poluan <pandu@××××××.info> wrote: |
5 |
>>>> |
6 |
>>>> On Dec 25, 2012 10:44 PM, "Mark Knecht" <markknecht@×××××.com> wrote: |
7 |
>>> <SNIP> |
8 |
>>>>> With the previous local drive I used ext3 and have had no problems. |
9 |
>>>>> I'm just wondering if there's a better choice & why. |
10 |
>>> <SNIP> |
11 |
>>>> |
12 |
>>>> For your usage, I think ext3 is the most suitable. |
13 |
>>>> |
14 |
>>>> Do you have another fs in mind? |
15 |
>>> |
16 |
>>> Really, no. ext3 has been fine. I didn't see any real advantage to |
17 |
>>> ext4 myself. Florian offers the removal argument but I've never |
18 |
>>> removed files from this database. It's just movies so the systems just |
19 |
>>> grows over time. |
20 |
>>> |
21 |
[...] |
22 |
> |
23 |
> I wonder if there's anything to be said for changing block sizes, etc. |
24 |
> away from whatever the defaults are? All of the files are currently |
25 |
> between 350MB & 1.2GB so there's never going to be many more than 2K |
26 |
> files on the drive and I'm assuming the rsync operation if file by |
27 |
> file so fragmentation in the beginning, and probably over time, is |
28 |
> going to be pretty low I think. |
29 |
> |
30 |
|
31 |
The default of 4k blocks is the largest possible and with big files |
32 |
there is no advantage in using smaller blocks. In fact, it could come |
33 |
back to bite you as it limits file sizes. I guess it also increases fsck |
34 |
times. |
35 |
|
36 |
Another issue I noticed is that it can cause terrible performance if the |
37 |
block size is smaller than the physical block size of the device. mke2fs |
38 |
warns when it detects this. |
39 |
|
40 |
Regards, |
41 |
Florian Philipp |