It seems as though that is the problem.  The last few
lines of output (in debug mode) are:

gmond: /dev/ksyms is not a 32-bit kernel namelist
kvm_open: Error 0
 *** WARNING!!!!  kvm_open() failed.  prepare for a
segfault ... *** 
*** kvm_open() failed, are you running gmond as root?
kvm_nlist: Bad address
Lookup up kstat:  km (unix?)='unix', ks
(system_misc?)='system_misc',kn (resulting
metric?)='ncpus'
ncpus: kc is 38518
ncpus: Just did kstat_lookup().
ncpus: Looked up.
ncpus: Kstat data type:  2, Value returned: 4, 4 4 4
ncpus: Kernel close.  Val returned: 4
metric_init: Assigning cpu_num value (4) to ncpus.
Segmentation fault

I'm trying to recompile it as 64-bit, but regardless
of the incantation (-m64, -mcpu=v9, -mcpu=ultrasparc,
-mptr64, or combinations of these) it gives me the
same problems.

I'm also trying some other compilers (e.g. the Sun
Workshop compiler) but they all have their unique
problems that prevent compilation at all.  I'm hacking
through that to see if one of them will compile
Ganglia instead of gcc.  However, if you have any tips
on how to compile it as a 64-bit binary, I would be
grateful.

Thanks,

Andy
[EMAIL PROTECTED]

--- Steven Wagner <[EMAIL PROTECTED]> wrote:
> Andrew Gill wrote:
> > I'm trying to get Ganglia to work on Solaris 8,
> and
> > seem to be hitting my head against a wall.  I can
> > compile it without any problems, using gcc-3.2.
> > 
> > However, the gmond binary exits immediately
> (return
> > code 0) and no gmond process runs in the
> background. 
> > A 'truss' of gmond doesn't show any errors.  The
> gstat
> > binary exits with a segmentation fault unless it
> is
> > run with the -h or -V options.
> 
> Didja run gmond with debug output on?
> 
> I'm pretty sure (I'd have to check) that my
> gmetric_init() code, as well as 
> my configure patch, screams bloody murder if it
> finds library mismatches. 
> In fact it should say, "kernel open failed, I am
> probably about to 
> segfault" if it's trying to open a 64-bit kernel
> with a 32-bit kopen().
> 
> None of my Solaris development was with gcc 3.2.  It
> was mostly with 2.95.2 
> and 3.0.1, on Solaris 8 (an E420R, in fact).
> 
> I *can* tell you that I have gmond running like a
> champ on about two dozen 
> such systems...
> 
> > Has anybody else encountered this situation? 
> Thanks
> > for your help.
> 
> No, but I'd sure like to hear more about it...
> 


__________________________________________________
Do you Yahoo!?
Faith Hill - Exclusive Performances, Videos & More
http://faith.yahoo.com

Reply via email to