Data Collection intervals irregular

[Igor Kiselev]

Hi,

Just noticed that Data Collection has inconsistent intervals for particular SNMP variables:

snmpColDump1
04/17/2003 04:57:09 AM switch1 5.56303e+06
04/17/2003 05:57:09 AM switch1 1.67504e+07
04/17/2003 07:57:09 AM switch1 1.71151e+07
04/17/2003 08:57:09 AM switch1 2.16759e+07

Why 06:57:09 polling is missing?

I checked data for other interfaces of the same switch1 for the same SNMP variable
and they did have data within 06:XX - 07:XX interval. It's quite intermittent.
Sometimes I see more than 1 polling for the specified interval:

snmpColDump2
02/15/2003 10:12:50 AM switch1 1.70141e+07
02/15/2003 11:12:49 AM switch1 1.1389e+07
02/15/2003 11:13:09 AM switch1 3.74034e+06
02/15/2003 12:13:08 PM switch1 1.03999e+07

It happens quite often, every 24 hours I see something like that.
Please help to understand, I assume that reports generated out of these files are inaccurate as well.


My SNMP expression is Computed by:
#
ifInOctets_60min "Non rate number of octets received over 1 hour.\nComputed by:\n (IfInputOctets * 3600)\nfor interface.".1.3.6.1.2.1.2.2.1.10. 3600 *
#

Igor

[Ron Kinner]

Just a guess since I know nothing of your "Data Collection" progam but SNMP normally uses UDP as its transport. UDP does not retransmit if a packet goes astray. It's up to the application to resend a UDP packet if it doesnot get the expected response within a reasonable period. Could be that you lost a packet (or several packets if it makes more than one attempt) and it does not bother logging the fact that the attempt failed. That would explain the first symptom and might mean you have a network problem (unless the target is the local host) but the second would appear to be your application's fault or perhaps whatever scheduling program is involved.

You might want to look at MRTG. It does a nice job and gives you pretty graphs too.

http://people.ee.ethz.ch/~oetiker/webtools/mrtg/

Ron