1 |
2018-02-08 20:13 GMT+02:00 Rich Freeman <rich0@g.o>: |
2 |
> On 08/02/18 19:11, gevisz wrote: |
3 |
>> |
4 |
>> I never used tmpfs for portage TMPDIR before and now decided to give it a |
5 |
>> try. |
6 |
>> |
7 |
>> I have 8GB of RAM and 12GB of swap on a separate partition. |
8 |
>> |
9 |
>> Do I correctly understood |
10 |
>> https://wiki.gentoo.org/wiki/Portage_TMPDIR_on_tmpfs |
11 |
>> that I can safely set in the fstab the size of my tmpfs to 12GB so |
12 |
>> that the chromium could be emerged in tmpfs (using the swap) |
13 |
>> without the need to set notmpfs.conf for chromium and the likes. |
14 |
> |
15 |
> You can try it, but for Chromium these days you might find that still |
16 |
> doesn't perform great. I have 16GB of RAM (no swap) and have moved |
17 |
> back to building on SSD for that one package (with ccache to help). |
18 |
> |
19 |
> In an ideal world swap would STILL be better than building on disk, |
20 |
> because it gives the kernel fewer constraints around what gets written |
21 |
> to disk. |
22 |
|
23 |
> Anything written to disk MUST end up on the disk within the dirty |
24 |
> writeback time limit. Anything written to tmpfs doesn't ever have to |
25 |
> end up on disk, and if it is swapped the kernel need not do it in any |
26 |
> particular timeframe. Also, the swapfile doesn't need the same kinds |
27 |
> of integrity features as a filesystem, which probably lowers the cost |
28 |
> of writes somewhat (if nothing else after a reboot there is no need to |
29 |
> run tmpreaper on it). |
30 |
|
31 |
> So, swapping SHOULD still be better than building on disk, because any |
32 |
> object file that doesn't end up being swapped is a saved disk IO, and |
33 |
> the stuff that does get swapped will hopefully get written at a more |
34 |
> opportune time vs forcing the kernel to stop what is doing after 30s |
35 |
> (by default) to make sure that something gets written no matter what |
36 |
> (if it wasn't deleted before then). |
37 |
|
38 |
Thank you for the reply. |
39 |
|
40 |
I probably try a pure tmpfs + swap solution. If it fails some day, I will |
41 |
then add notmpfs exceptions. |
42 |
|
43 |
However, it probably won't be sooner than |
44 |
# emerge --update --deep --with-bdeps=y --newuse --backtrack=90 --ask |
45 |
world --exclude chromium |
46 |
fails because of the "--exclude chromium" part :), as I have already compiled |
47 |
the recent vertion of chromium with /var/tmp/portage on the hard disk and |
48 |
it took more than 24 hours on my old AMD Athlon X2 with j2 option. :( |