Net plugin bytes sent/recieved resets to 0 if host reboots

[douglasg14b]

Bug? Maybe feature request?

On freebsd.

Current behavior:

Collecting Net data bytes_sent and bytes_recv go sup over time, as expected. However when the host reboots these are set to 0 and go up again as data is sent/recieved.

Desired behavior:

The bytes_sent and bytes_recv maintain their previous values and continue to go up.

Use case:

When this reset to 0 is completely breaks any total bandwidth usage queries as it relies on the first vs last bytes_sent and bytes_recv for the timerange to get a total.

Ie. start: 100 bytes, end: 4000 bytes. Total for range is 3900 bytes. If it was to reset to 0 in the middle of a daterange you get negative bandwith usage using this.

[voiprodrigo]

It’s the OS that resets the counter. Use non negative difference or non negative derivative for calculations.

_{Sent with GitHawk}

[douglasg14b]

@voiprodrigo Can you explain what you mean by a non-negative derivative to avoid something like this. Do you have an example? I've been struggling to figure this out for weeks, and have not been able to find an answer.... I'm pretty desperate at this point.

I'm not using a derivative for this item

Example of what happens when it sets to 0 in the middle of a time range:

Query:

SELECT (last("bytes_sent") - first("bytes_sent")) + (last("bytes_recv") - first("bytes_recv")) FROM "net" WHERE ("interface" = 'em4') AND $timeFilter GROUP BY time($__interval) fill(null)

[douglasg14b]

Oh I see, I can use SELECT non_negative_difference("bytes_recv") FROM "net" WHERE ("interface" = 'em4') AND $timeFilter to get the difference.

Though it does make sparklines look terrible.

[voiprodrigo]

Add a last() to bytes_recv within non_negative_difference() and group by time. The larger the time value, the less resolution you'll see in the lines. Suit to taste.