1 |
2008/9/18 Ricardo Saffi Marques <saffi@××××××××××××××.br>: |
2 |
> |
3 |
> In my case, my cron runs "eix-sync" every night at 00h and I just did an "time emerge -DNuvp world" |
4 |
> here, getting: |
5 |
> |
6 |
> Total: 129 packages (103 upgrades, 18 new, 8 reinstalls), Size of downloads: 187,761 kB |
7 |
> |
8 |
> real 0m21.074s |
9 |
> user 0m18.913s |
10 |
> sys 0m0.724s |
11 |
> |
12 |
> So I assume once a day (let's say, every night) is a good choice for syncing, right? :-) |
13 |
> Never ever had a problem with slow dependency calculation on emerge. |
14 |
|
15 |
As far as I know the rsync servers are syncing against the master |
16 |
server(s) about every half an hour which obviously also updates the |
17 |
timestamp. So what I described above only happens if you sync within a |
18 |
shorter period. Running eix-sync every day shouldn't be a problem. |
19 |
|
20 |
Syncing twice an hour is also considered bad practise because of the |
21 |
heavy traffic involved. It just happens to me sometimes if I test |
22 |
stuff and had to sync within a shorter period. |
23 |
|
24 |
I just asked this because it could be possible that the script calls |
25 |
eix-sync twice or after an emerge --sync which is not necessary or for |
26 |
some other reason it is called twice within a too short period. |
27 |
|
28 |
Regarding the fragmentation. I use ext3 and reiserfs for the portage |
29 |
tree and did not recognize a significant slowdown. There is a slowdown |
30 |
but not from 4 to 20 minutes. At least in my case. |
31 |
|
32 |
Regards, |
33 |
|
34 |
Daniel |