On Tue, 2012-04-03 at 15:30 -0400, Lomax, Michael P. wrote: > Again my getting your response did not happen due to a problem with my > email. I downloaded your text below… > > The files you requested are attached. > > Your assistance is much appreciated… > > > Mike Lomax > On Mon, 2012-04-02 at 16:52 -0400, Lomax, Michael P. wrote: > > You may recall that I had asked for your assistance back in August > > with an attempt to load Net-SNMP 5.7.1.pre1 on several AIX 6.1 > > servers. Unfortunately I got pulled away from my project back then > > and am now, finally, getting back to it. > > > > This time I started with the Net-SNMP 5.7.1 release using the tar.gz > > file from your download page. The `configure` and `make` were > > successful but I did run into a problem with the `make test`. All > > tests passed with the exception of T0072com2secunix_simple. This > > particular test produced errors which can be found in the attached > > snmpd.log file taken from the /tmp/T0072com2secunix_simple* > directory. > > T072 is supposed to do quite a few negative tests so that you get > errors > is expected, but looking at the errors you get is somewhat suspicious > as > I would expect the errors to match the expectations in the second part > of T072com2secunix_simple. > > > > Looking back at my attempts in August I found that these same errors > > existed during those runs as well. Is this something that is a > known > > problem? Any suggestions? > > As I say, some of them are expected but not all. > > Could you please send T0072com2secunix_simple > and /tmp/T0072com2secunix_simple*/snmpd.conf
Thanks. The test that fail are, principally, tu03c and tu03d but they fail in interesting ways. It seems that on AIX the sun_path member of sockaddr_un is 1022 bytes, could you please verify this by checking in /sys/un.h and preferably also by checking that there is a line in include/net-snmp-config.h that reads #define SIZEOF_SOCKADDR_UN_SUN_PATH 1022 Now, it seems that our config reading routines have a line length maximum of 1024 bytes and that the overflow handling i less than graceful but this is still only a theory so I will need to check this further. The bug seems to appear in the intersection of these limits. /MF ------------------------------------------------------------------------------ Better than sec? Nothing is better than sec when it comes to monitoring Big Data applications. Try Boundary one-second resolution app monitoring today. Free. http://p.sf.net/sfu/Boundary-dev2dev _______________________________________________ Net-snmp-coders mailing list Net-snmp-coders@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/net-snmp-coders