1 |
On Wed, Jan 7, 2015 at 3:17 PM, Neil Bothwick <neil@××××××××××.uk> wrote: |
2 |
> On Wed, 7 Jan 2015 14:25:52 -0500, Rich Freeman wrote: |
3 |
> |
4 |
>> > Not as a general FS, but as a specific choice for $PORTAGE_TMPDIR it |
5 |
>> > may be worth testing. XFS was designed for an environment that used |
6 |
>> > temporary files that didn't need to be committed to disk, so its |
7 |
>> > caching doesn't write to disk anywhere near as often. That means you |
8 |
>> > would be working with files stored in RAM a lot of the time. |
9 |
>> > |
10 |
>> |
11 |
>> If you're going to be saving the build files using mv/ln (or with cp |
12 |
>> --reflink on a filesystem that supports this), then the LAST thing I |
13 |
>> would do is use a different fs for PORTAGE_TMPDIR. The |
14 |
>> best-performing option would be to make it a directory on the same |
15 |
>> filesystem as wherever you're going to store the files by |
16 |
>> moving/(ref)linking them. That makes the move/(ref)link operation |
17 |
>> require minimal IO. |
18 |
> |
19 |
> That's not what I meant, but I see your point about using --reflink if |
20 |
> making copies. My thought was to forget the whole tmpfs and copying |
21 |
> think, set KEEP WORK in FEATURES and use XFS for PORTAGE_TMPDIR. That way |
22 |
> most of the work is taking place in the cache without the frequent disk |
23 |
> writes of other filesystems. I would expect XFS to be faster for this |
24 |
> job, but have no data to support that assumption. |
25 |
> |
26 |
> Or you could just put TMPDIR on btrfs and snapshot after each emerge. |
27 |
> |
28 |
|
29 |
If you're not going to move it but just leave it all in TMPDIR |
30 |
forever, then by all means use whatever filesystem you like. Sure, |
31 |
you could snapshot it and clean it, but that might not be the cleanest |
32 |
way. |
33 |
|
34 |
-- |
35 |
Rich |