1 |
Neil Bothwick <neil <at> digimed.co.uk> writes: |
2 |
|
3 |
> |
4 |
> |
5 |
> On 7 January 2015 13:16:03 GMT+00:00, Rich Freeman <rich0 <at> gentoo.org> |
6 |
wrote: |
7 |
> On Wed, Jan 7, 2015 at 3:36 AM, Neil Bothwick <neil <at> digimed.co.uk> wrote: |
8 |
> Personally, I wouldn't bother, there is not that much of a gain when |
9 |
using tmpfs, so if you want to keep all the working files after compilation, |
10 |
the extra overhead and complexity of copying to hard disk would make it not |
11 |
worthwhile. Just stick to a spinning disk. |
12 |
|
13 |
Hmmmm. |
14 |
|
15 |
|
16 |
> I've found that compiling on tmpfs has a fairly significant performance |
17 |
gain, but you'd completely negate it of you copied everything to a hard |
18 |
drive anyway.When you build on a hard drive the file system is going to |
19 |
treat all those intermediate object files with great care and ensure that |
20 |
they aren't lost in the event of a power failure, forcing them to be |
21 |
committed to disk within 30 seconds (typically) and blocking IO |
22 |
|
23 |
OK. |
24 |
|
25 |
> when that happens. Then a minute later it is going to go and have to |
26 |
delete all those files it so carefully committed. When you build on tmpfs |
27 |
nothing gets written to your file system except for the final output of the |
28 |
build. All those intermediate files are deleted having never blocked your |
29 |
disk IO. However, if your goal is to actually save them anyway, then you |
30 |
might as well just build on disk - moving files doesn't take all that much |
31 |
IO on most file systems.If your goal is to create tarballs of the saved |
32 |
builds then you're probably still better off building on tmpfs and creating |
33 |
the tarball via a hook or something. |
34 |
|
35 |
Yea, I could not rationalize a solution either, hence the post. |
36 |
|
37 |
|
38 |
> That's why I thought XFS may help. |
39 |
> Reports of the speed gain from tmpfs are quite mixed, but I do use it myself. |
40 |
|
41 |
I'm moving to btrfs and eventually ceph, so xfs is not on my roadmap.... |
42 |
|
43 |
I do not have performance problems (FX8250 with 32G ram). And eventually |
44 |
I'll stumble into something robust with clustering. (mesos-distcc)[1] and |
45 |
others efforts will yield what I desire (eventually) or something very fast |
46 |
and close enough [2]. |
47 |
|
48 |
|
49 |
I had some overall resistance to the idea too, but I thought it best to |
50 |
query the list to see if something robust and simple did already exist.... |
51 |
|
52 |
|
53 |
Thanks for all the posts, |
54 |
|
55 |
James |
56 |
|
57 |
|
58 |
[1] https://github.com/mesos/mesos-distcc |
59 |
|
60 |
[2] http://www.zentoo.org/ (Continuous Integration) |