| 1 |
On Wed, 07 Jan 2015 10:00:17 +0800, Andrew Lowe wrote: |
| 2 |
|
| 3 |
> # |
| 4 |
> # RAM disk for emerges |
| 5 |
> # |
| 6 |
> tmpfs /var/tmp/portage tmpfs |
| 7 |
> uid=portage,gid=portage,mode=0775,size=8192M,noatime 0 0 |
| 8 |
> |
| 9 |
> |
| 10 |
> The above is all ONE line. This has worked fine for me. I have a machine |
| 11 |
> with 16GB so I decided to go half/half on the RAM disk hence the 8GB |
| 12 |
> size and tune if need be. I've seen no problems hence have left it at |
| 13 |
> that. |
| 14 |
|
| 15 |
size efaults to 50% for tmpfs so setting it to 8G is unnecessary. |
| 16 |
|
| 17 |
> > Second question: |
| 18 |
> |
| 19 |
> No idea :) Would some sort of cp/rsync running under a short |
| 20 |
> interval cron job do what you want. As I said, I really have no idea |
| 21 |
> here. |
| 22 |
|
| 23 |
You could define a post_src_compile() in /etc/portage to make the copy |
| 24 |
when the compile completes. You would also need to register a fail hook |
| 25 |
to do the same when the compile fails. |
| 26 |
|
| 27 |
Personally, I wouldn't bother, there is not that much of a gain when |
| 28 |
using tmpfs, so if you want to keep all the working files after |
| 29 |
compilation, the extra overhead and complexity of copying to hard disk |
| 30 |
would make it not worthwhile. Just stick to a spinning disk. You may find |
| 31 |
that a filesystem like XFS with its aggressive caching, would give a |
| 32 |
similar performance. |
| 33 |
|
| 34 |
|
| 35 |
-- |
| 36 |
Neil Bothwick |
| 37 |
|
| 38 |
God is real, unless specifically declared integer. |
Archives