Interpretating values for Network statistics


I’ve downloaded and installed the Percona Linux template. I noticed the values for tx-drop had values of ‘m’ and ‘u’ and don’t quite understand why I’m getting values of micro and milli.

I was expecting reading to simply be the number of packet dropped at that point of time but it doesn’t appear so.

So I’m interpreting it as follows:

  • A defined time period divided by the number of dropped packets. eg. 12.5m would mean 80 packets dropped within a second. 150u would mean 6666.67 packets dropped within a second (1/150u)

Is this interpretation of the reading correct? Thanks.

TX-drop is constantly increasing counter (derive in Cacti).
So the daily graph (5 min avg) should reflect a rate of change in packets dropped/sec within 5 min. interval.

The current value on the graph (for derive) is calculated as (prev_value - cur_value) / 300.
If you see 12.5m, it should translate to X / 300 = 12.5m. Therefore, X = 300 * 12.5/1000 = 3.75 packets dropped within the last 5 min.
Accordingly, 150u == 0.045 packets.

Of course, it shouldn’t be decimal but may be you were looking to weekly or monthly graph where interval is not 300?

To verify:
[root@host ~]# cat /proc/net/dev
Inter-| Receive | Transmit
face |bytes packets errs drop fifo frame compressed multicast|bytes packets errs drop fifo colls carrier compressed
lo:1592496297 2435686 0 0 0 0 0 0 1592496297 2435686 0 0 0 0 0 0
eth0:6563229868651 4662142281 0 0 0 0 0 66380 354572077461 2642132444 0 164 0 0 0 0

Capture “drop” number at the full 5-min. time, i.e. at XX:X5:00. Wait 5 min. and get the new “drop” number.
Calculate the change as (prev_value - cur_value) / 300 and compare with what is shown on the daily graph.