1 |
On 07/01/18 20:55, Ian Zimmerman wrote: |
2 |
> |
3 |
> I ought to disclose that the server is Debian. But the distcc versions |
4 |
> on both sides were the same, and I hand-compiled a matching gcc version |
5 |
> on the server. |
6 |
> |
7 |
> One thing I very much dislike about distcc is that there seems to be no |
8 |
> good way of using a full path to the compiler on the server that's |
9 |
> different from the one on the client. |
10 |
> |
11 |
|
12 |
I think that's probably the cause of the problem you've been having. I'm |
13 |
no distcc expert by any means but you need to have binutils/libtool/gcc |
14 |
identical across all machines, and in the case of different distros it |
15 |
means hand compiling these on every machine, not just one - distros tend |
16 |
to include their own fixes. |
17 |
|
18 |
I found this article: |
19 |
|
20 |
http://rostedt.homelinux.com/distcc/ |
21 |
|
22 |
which details some of these problems, but in that example they were also |
23 |
compiling x32 and x64. |
24 |
|
25 |
I've never had a problem with distcc in my all-Gentoo environment, other |
26 |
than some random packages that don't like distcc. |
27 |
|
28 |
I was actually reading about pump mode and considering trying that on my |
29 |
Celerons. |
30 |
|
31 |
Dan |