1 |
http://code.google.com/p/bashttpd/ |
2 |
|
3 |
run with systemd or xinetd |
4 |
|
5 |
|
6 |
|
7 |
于 2011年11月14日 18:05, J. Roeleveld 写道: |
8 |
> On Sat, November 12, 2011 2:11 pm, YoYo Siska wrote: |
9 |
>> On Sat, Nov 12, 2011 at 07:40:08PM +0700, Pandu Poluan wrote: |
10 |
>>> During my drive home, something hit my brain: why not have the 'master' |
11 |
>>> server share the distfiles dir via NFS? |
12 |
>>> |
13 |
>>> So, the question now becomes: what's the drawback/benefit of NFS-sharing |
14 |
>>> vs |
15 |
>>> HTTP-sharing? The scenario is back-end LAN at the office, thus, a |
16 |
>>> trusted |
17 |
>>> network by definition. |
18 |
>> |
19 |
>> NFS doesn't like when it looses connection to the server. The only |
20 |
>> problems I had ever with NFS were because I forgot to unmout it before a |
21 |
>> server restart or when I took a computer (laptop) off to another |
22 |
>> network... |
23 |
> |
24 |
> NFS-shares can work, but these need to be umounted before network goes. |
25 |
> If server goes, problems can occur there as well. |
26 |
> But that is true with any server/client filesharing. (CIFS/Samba, for |
27 |
> instance) |
28 |
> |
29 |
>> Otherwise it works well, esp. when mounted ro on the clients, however |
30 |
>> for distfiles it might make sense to allow the clients download and save |
31 |
>> tarballs that are not there yet ;), though I never used it with many |
32 |
>> computer emerging/downloading same same stuff, so can't say if locking |
33 |
>> etc works correctly... |
34 |
> |
35 |
> Locking works correctly, have had 5 machines share the same NFS-shared |
36 |
> distfiles and all downloading the source-files. |
37 |
> |
38 |
>> And with NFS the clients won't duplicate the files in their own |
39 |
>> distfiles directories ;) |
40 |
> |
41 |
> Big plus, for me :) |
42 |
> |
43 |
> -- |
44 |
> Joost |
45 |
> |
46 |
> |