1 |
Andrew D. Fant wrote: |
2 |
> I'll dig around and forward the ebuild to the list later tonight or |
3 |
> tomorrow. At the risk of going off on a rant, I think that the |
4 |
> /etc/env.d model is a core part of the problem. If two users want to |
5 |
> use two different mpi implementations, will eselect allow them to both |
6 |
> use them at the same time, or will it have to be one or the other at |
7 |
> once? As I see it, something will have to be tweaking $LD_LIBRARY_PATH |
8 |
> for each user's environment, |
9 |
|
10 |
Yep, that's the plan. Some things just won't work as well because they |
11 |
don't really respect variables, like the includes. |
12 |
|
13 |
Something like this: |
14 |
|
15 |
MPI_IMPLEMENTATION="$implementation" |
16 |
PATH="/usr/lib/mpi/$implementation/$version/$compiler/bin:$PATH" |
17 |
LD_LIBRARY_PATH="/usr/lib/mpi/$implementation/$version/$compiler/lib:$LD_LIBRARY_PATH" |
18 |
CPPFLAGS="-I/usr/lib/mpi/$implementation/$version/$compiler/include:$CPPFLAGS" |
19 |
MANPATH="/usr/lib/mpi/$implementation/$version/$compiler/share/man:$MANPATH" |
20 |
INFOPATH="/usr/lib/mpi/$implementation/$version/$compiler/share/info:$INFOPATH" |
21 |
|
22 |
> or /etc/ld.so.cache will have to be changed |
23 |
> for each user's login to reflect their prefered mpi environment. |
24 |
|
25 |
That totally screws with multiple simultaneous logins. Environment |
26 |
variables will have to be the way to go. |
27 |
|
28 |
> $PATH |
29 |
> and $MANPATH can be handled by judicious checks in /etc/csh.cshrc and |
30 |
> /etc/profile, but since /etc/env.d insists on manipulating the shared |
31 |
> object cache directly, this has the potential to get really ugly without |
32 |
> some cooperation from the core system team. I don't think that the |
33 |
> cluster herd can fix it on it's own. |
34 |
|
35 |
I think we can, but I could certainly be proven wrong on this. |
36 |
|
37 |
> I would love to be proven wrong and will gladly defer to anyone who does |
38 |
> so, I just know that I've been trying to outthink this problem for a |
39 |
> year or so on our cluster and haven't come up with a sane solution for a |
40 |
> true multi-user environment. |
41 |
|
42 |
I would love to find a way that works too. =) |
43 |
|
44 |
Thanks, |
45 |
Donnie |