Re: [Freeipmi-users] ipmiping adversely affecting ipmitool

[Albert Chu]

Hi Dick,

On Sun, 2010-08-01 at 16:13 -0700, address@hidden wrote:
> Hello all,
> 
> I am working on a multi-threaded app that searches a range of a IP
> addresses for IPMI devices and gathers info about any IPMI devices it
> finds.

Just as an FYI, you're aware of FreeIPMI's 'ipmidetect' tool and
daemon??  Not sure if that would be suitable for your needs.

> A thread is spun for each IP address to be searched.  I am using
> ipmiping from the freeipmi suite to find them and ipmitool to get the
> info.  I have isolated an interesting problem.  If a thread for an IP
> address that will fail the ipmiping is included in the range, than
> other threads will from time to time fail getting information via
> ipmitool.  The symptom is that the ipmitool request will return an OK
> status but there will be no data returned.

If each thread is given 1 IP address to handle, but 1 thread's failure
is affecting other threads, I'm not sure how this would be issue with
ipmiping.  Appears there is a threading issue.

> You could test this by setting up an easy program that spins threads
> that perform ipmiping, then if 1.5 or 2.0 are returned, send an fru
> print command using ipmitool, then echo or save the output to a file.
> 
> I wonder if ipmiping is leaving the open-ipmi driver in a weird state
> so that when ipmitool runs its confused at times.

I'm not sure why the open-ipmi kernel driver is involved here.  If the
ipmiping succeeds, are you doing the 'ipmitool' in-band?  Anything
out-of-band should not affect the kernel driver.

> Any ideas what this might be?

I can only gather one guess.  Some BMCs I've seen can get "confused" if
they receive too many duplicate packets with duplicate sequence numbers
and a short time span (DOS prevention?  bug in firmware?  BMC slow and
can't handle so many packets simultaneously?  Your guess is as good as
mine).  Depending on how you're doing the ipmiping or ipmitool, that
could be an issue.

Al

> Thanks,
> 
> Dick Detweiler |Senior Software Engineer, MSD | Avocent | USA
> Emerson Network Power | 4991 Corporate Drive | Huntsville, AL 35805-6201
> T 256.261.6550 | F 256.430.4027
> www.*avocent.com<http://*www.*avocent.com>
> www.*emersonnetworkpower.com
> 
> _______________________________________________
> Freeipmi-users mailing list
> address@hidden
> http://*lists.gnu.org/mailman/listinfo/freeipmi-users
> 
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory