SNMP Discovery not receiving all OIDs from F5s

Jason DaSilva
Giga Expert

Good day to all,

Please don't confuse this with issues with a missing OID for classification.  We have 20 F5 BIG-IP load balancers in one of our test discovery schedules.  I can run that schedule back to back (5 minutes to let the first finish) and in the first run 1-3 devices fail to classify, and in the second run a different 1-3 will fail to classify (often different from the first, but the ones that failed the first time may pass the second time).
Looking over the XML data in one specific case, the failed pass returned 49 OIDs, the classified run returned 52 OIDs.  I am positive that I see where the fail is:

JasonDaSilva_1-1681308827402.png

 

In this image, even though the <sysDescr> line fills in the text for the device type, the actual OID line is missing.

The only think I can think of is that it's timing out on receiving that line?  SNMP is not one of my strong areas.  Just curious if this is a common issue forcing us to run our F5 discoveries multiple times in a row to get proper data, or if there is something I can tweak to make sure that these devices are discovering properly on the first pass.

 

Thanks.

 

1 ACCEPTED SOLUTION

Jason DaSilva
Giga Expert

Poking around a bit more, I ended up finding the fix.  On each MID server that is involved with SNMP discovery, you would need to add a new configuration parameter (shown in the tabs under related links).  This parameter is called mid.snmp.request.timeout.  It defaults to 1500ms, and for me changing it to 2000ms fixed my problems.

View solution in original post

3 REPLIES 3

Jason DaSilva
Giga Expert

Poking around a bit more, I ended up finding the fix.  On each MID server that is involved with SNMP discovery, you would need to add a new configuration parameter (shown in the tabs under related links).  This parameter is called mid.snmp.request.timeout.  It defaults to 1500ms, and for me changing it to 2000ms fixed my problems.

Was about to come in here and suggest the same thing you found yourself; adjust the timeout.

 

I was doing something similar previously with some UPS devices and their response was crazy slow, so much so I had to test on a "dedicated" MID Server.

 

Thanks for checking in.  The only down side is that it has to be done on all MIDs that are running SNMP discovery.  Right now, in our DEV, we have 2, but once we hit production, we have about 9 that will be doing the general discovery.