Hello, all,
In my company, we use a multi-threaded tool to perform SNMPGET
requests on many different devices at various times. From time to
time, we have experienced a strange problem where, all of the sudden,
our tool will get back some failure condition (not a timeout; other
than that, we're not sure yet), and *every* OID that is polled after
that point, for any device (even with new sessions created), will fail
in the same mysterious manner.
Interestingly enough, we found that at a particular time when this
occured (and it may occur once per month if we are extremely lucky to
catch it, with this tool doing millions of requests) that *all* of the
devices with a *particular* community string (alphanumeric; maybe 12
characters, tops) failed. Ones with a string of "public" did not
succumb to the problem. In *all* cases, proper requests are coming
out of our box, and proper responses are coming back to it (tcpdump
confirms this).
Now, this has been plaguing us for *years* and we have simply coded
around it by building in some global variables that count failures and
restart our tool if that number exceeds a certain number in a row
(which is horrible, but it does the trick).
We are currently using version 5.4.1 (gentoo-r3), and I have not been
able to find a bug that looks similar. Our code has a function that
looks a little something like the following; essentially, we are given
a list of OIDs that we want to get at once. We create a PDU, pack it,
and then call "snmp_sess_synch_response" on it. Nothing crazy, and it
works all of the time (except the random times that it does not).
int snmpGetMany( ... ) {
...
pdu = snmp_pdu_create( SNMP_MSG_GET );
for( i = 0; i < oidsToGetSize; i++ ) {
snmp_add_null_var( pdu, oidsToGet[i], oidSizes[i] );
}
status = snmp_sess_synch_response( sessionHandle, pdu, &response );
...
}
If anyone here has *any* insight whatsoever into this, we have been
looking (halfheartedly) for years and have no stepped up our search
again, running full debug on select environments in attempt to catch
this happening and see if we can figure out what's going on. If you
can help in any way, please let me know.
Thanks,
Doug
------------------------------------------------------------------------------
Beautiful is writing same markup. Internet Explorer 9 supports
standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 & L3.
Spend less time writing and rewriting code and more time creating great
experiences on the web. Be a part of the beta today.
http://p.sf.net/sfu/beautyoftheweb
_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
