[ofa-general] Re: [RFC][PATCH 0/4] opensm: using conventional config file

Ira Weiny weiny2 at llnl.gov
Mon May 12 14:45:41 PDT 2008


Sasha,

Also, I wonder if anyone would object to you applying your patches to the tree
as is and we work out the details from there?  I don't see anything wrong with
your patches except that more work will be needed, as you said, in the man
pages and scripts.

After you apply your patches I think we can start in changing the man pages and
scripts.  Al, Tim, and I started talking about this after I ran into problems
with the current config files trying to write a PerfMgr HOWTO today.  :-(

Ira

On Mon, 12 May 2008 11:33:45 -0700
Al Chu <chu11 at llnl.gov> wrote:

> Hey Sasha,
> 
> Ira and I were chatting.  A few other comments:
> 
> 1) Many configuration values are not output by default in opensm right
> now, mainly b/c it behaves like a cache rather than an configuration
> file.  i.e.
> 
>         if (p_opts->connect_roots)
>                 fprintf(opts_file,
>                         "# Connect roots (use FALSE if unsure)\n"
>                         "connect_roots %s\n\n",
>                         p_opts->connect_roots ? "TRUE" : "FALSE");
> 
> Going forward w/ a config file, I think these should be output by
> default all the time so users know they exist.
> 
> 2) Will there be an option to specify an alternate configuration file,
> i.e. not /etc/opensm/opensm.conf?
> 
> Al
> 
> On Wed, 2008-04-09 at 01:10 +0000, Sasha Khapyorsky wrote:
> > Hi,
> > 
> > This is attempt to make some order with OpenSM configuration. Now it
> > will use conventional (similar to another programs which may have
> > configuration) config ($sysconfig/etc/opensm/opensm.conf) file instead
> > of option cache file. Config file for some startup scripts should go
> > away. Option '-c' is preserved - it can be useful for config file
> > template generation, but OpenSM will not try to read option cache file.
> > 
> > This is RFC yet. In addition to this we will need to update scripts and
> > man pages.
> > 
> > Any feedback? Thoughts?
> > 
> > Sasha
> -- 
> Albert Chu
> chu11 at llnl.gov
> 925-422-5311
> Computer Scientist
> High Performance Systems Division
> Lawrence Livermore National Laboratory
> 



More information about the general mailing list