>>>>> On Fri, 1 Oct 2004 10:55:59 -0700, "Jeff Johnson" <[EMAIL PROTECTED]> said:

Jeff> As I have previously mentioned, I'm trying to convert our
Jeff> 5.0.7-based code to 5.1.2.  One of the things I've encountered
Jeff> is a fix to a bug that was applied locally, but apparently was
Jeff> never communicated to the net-snmp group, or was never applied,
Jeff> since the bug is present in the
Jeff> 5.1.2 code.  I've just opened bug 1038616 against this problem, which
Jeff> also seems related to bugs 860414 and 865415.  The bug description
Jeff> follows.  A patch has been attached to the bug.

Out of curiosity, have you tried the 5.2.pre1 code as I *thought* that
problem had been fixed...  May be wrong though.

Jeff> At this point, if you query the interfaces MIB, you'll discover
Jeff> that all of the wlan0 objects are now using instance 2 instead
Jeff> of instance 3.  This is a violation of the MIB, as well as
Jeff> something that just isn't right.

FYI, though true managers are still out of luck.  There is no
requirement in the MIB that indexes be constant from one boot to the
next.  Thus the only thing you can be sure of is when the sysUpTime
rolls back to zero you need to re-check for the interface indexes
again anyway.  Don't get me wrong, our agent shouldn't be doing what
it does (and always has).  But don't ever write manager code that
assumes indexes won't change, because they will across a restart.

In the net-policy GUI for making graphs (which I wrote) I actually ask
the user what they want to use as an index to the graph.  Thus you can
say "graph this host's incoming bandwidth usage based on the value of
ifName = eth0" rather than having to rely on ifIndex = 1 and not
knowing what that means.  Much better.

Jeff> In order to fix this, the interfaces MIB code must be modified to use
Jeff> the SIOCGIFINDEX values instead of the ordinal order in /proc/nt/dev
Jeff> when determining the ifIndex of a given interface.

I believe the fix that I thought Dave? wrote actually cached the names
and used previously cached numbers based on interface names.  May be
wrong though.



-- 
Wes Hardaker
Sparta


-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Net-snmp-coders mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders

Reply via email to