In the spirit of answering my own question from a few weeks back...

"Tackabury, Wayne" <[email protected]> axxxed....

>I'm trying to compile in support for the 64-bit ifXTable counter support for 
>the IF-MIB for Solaris 10 Intel
>(and Sparc ultimately)...If I reestablish the build environment with 
>-enable-mfd-rewrites specified, I get an
>agent that starts up printing over and over to stdout
>
>            insert (-1) error on subcontainer 'interface container'

I actually figured out the issue here.  Somehow, an earlier mail (that I 
discovered a bit later) from a chap I definitely owe a beer or three to named 
Andrew Cosgriff noted:

>Yes, I have the same problem if I compile 5.4.2.1 with
>--with-mfd-rewrites and try running it in a solaris zone.

(this from 
http://groups.google.com/group/mailing.unix.net-snmp-users/browse_thread/thread/f143f3297bd7eeda)
 -- the organization of the SF archives makes digging this up here a bit 
difficult.

D'oh!  It makes a lot of sense, although it doesn't make sense why the 
enable-mfd-rewrites support level would fail at the level of kernel 
virtualization provided in zoned systems, but work for non-mfd/32-bit counters. 
 I'm guessing that either a) the implementation would actually work if the 
agent was running in the zone master and this is in fact an issue of some 
kernel call made that doesn't get back the response expected or accesses a 
kernel structure unsupported on zone suboridinates or, b) this is just a dumb 
bug having to do with presumptions about what the interface names will be on 
Solaris that gets blown up with the way zone-specific if-names are decorated--a 
simple regex problem, if you will.

I've not yet looked into which of those two it is, but I've worked around the 
issue for now (working fine on a non-zoned system, and that fallback is buying 
me some time).  If in fact the issue is (a) above, which means it's a somewhat 
interesting issue of agent placement, occurs to me this could be one of a set 
of issues that might be fodder for a "running net-snmp in virtual machine 
environments" FAQ.

Anyways, should anybody else run into this and have their search filter find 
this response before Andrew's, that's the issue, at least the one I encoutered.

Regards,
Wayne Tackabury
Mirror Image Internet

P.S.: Andrew Cosgriff...if you're out there, thanks a bunch and I'm serious 
about the brews.

------------------------------------------------------------------------------
_______________________________________________
Net-snmp-users mailing list
[email protected]
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users

Reply via email to