Feature suggestions: Average of duration and additional Logic

Vandabee · 2018-04-16 11:30

Hi, this is a suggestion to improve the average load flexibility which was implemented in topic #2374 - Average CPU and Network Transfer.

If the time length over which to measure the average could be set independently of the duration criteria, that would be great.

For instance I wish to suspend the pc if the network load has been below 500 KB/S for 10 minutes. If I could set a custom average of for instance 30 seconds, it would even out spikes.
With the average going over the entire duration period, it takes a full duration from the network activity stops, until the network traffic average falls below the threshold. With the current version, the alternative is to not use the average function, in which case the timer is reset with each spike.

Complementary suggestion:
An additional logic, could be used to achieve the same; “1BY1-resume”. In this logic events should still be monitored after they are passed. If an already passed event exceeds the threshold, the logic should always return to the earliest event which is no longer true, or start at the top if that is easier to code..
In this case, I could set a Network threshold of 30 seconds with average, followed by a ten-minute count down. The logic would be reset each time the network activity goes above the threshold. In this way I can suspend the pc after ten minutes of inactivity.

Thanks for the great application!

Last edited by Vandabee (2018-04-17 08:06)

den4b · 2018-05-13 12:55

Vandabee wrote:

If the time length over which to measure the average could be set independently of the duration criteria.

A change in the sampling rate should not affect the final average reading per second, so I don't see why would you need this option.

Vandabee wrote:

For instance I wish to suspend the pc if the network load has been below 500 KB/s for 10 minutes. If I could set a custom average of for instance 30 seconds, it would even out spikes.

Again, the sampling rate of 30 seconds vs 1 second will both produce the same average KB/s reading in 10 minutes.

Perhaps I did not understand your request correctly?

Vandabee · 2018-05-13 16:22

I am fine with the sampling rate, for a ten minute periode, I am just looking for the running average of for instance 30 samples instead of 600.

I am looking for the running average of X samples to be beneath the threshold for Y seconds.

I need for the usage to be beneath the threshold for ten minutes.
If I use average, it takes the average over ten minutes. After a period of high usage it takes almost 10 minutes for the average to fall below the threshold, after which the counting begins. In this senario it takes ~20 minutes for event to trigger.
Alternatively I don't use the average function, in which case the counting restarts everytime there is a usage spike of a few seconds.

If I could set the period length for which the average is taken independently from the monitotring period, then it wouldn't have to take a full monitoring period for the average to fall beneath the threshold and the countfown wouldn't be reset with each usage spike.

Last edited by Vandabee (2018-05-13 16:29)

den4b · 2018-05-14 13:24

Vandabee wrote:

I am fine with the sampling rate, for a ten minute period, I am just looking for the running average of for instance 30 samples instead of 600.

Then why not set the monitoring period to 30 seconds, i.e. below 500 KB/s average for 30 seconds?

You can also have several simulation events with different thresholds across different periods with a continuous "ALL" logic, for example:

1) 500 KB/s average for 30 seconds.
2) 100 KB/s average for 10 minutes.

This kind of allows for short recent bursts, but a relatively low traffic over longer period.

Vandabee · 2018-05-14 13:51

If usage has been high for 10 minutes+, then it will take ten minutes before 2) starts counting, which will add another 10 minutes.

This would give me 10 minute timeout after low usage, 20 minute timeout after high usage. Not the end of the world, it is the way I already have it setup.

den4b · 2018-05-15 01:04

Essentially, Shutter needs a more fine grained logic controls, which would allow for more complex workflows.

This feature is at the brainstorming stage, where we need to strike the right balance between flexibility and ease of use.

Feedback and suggestions are always welcomed!

Vandabee · 2018-05-15 07:35

I have made a suggestion for that in "Advanced logic for events". It is quite simple and very versatile.

den4b Forum

#1 2018-04-16 11:30

Feature suggestions: Average of duration and additional Logic

#2 2018-05-13 12:55

Re: Feature suggestions: Average of duration and additional Logic

#3 2018-05-13 16:22

Re: Feature suggestions: Average of duration and additional Logic

#4 2018-05-14 13:24

Re: Feature suggestions: Average of duration and additional Logic

#5 2018-05-14 13:51

Re: Feature suggestions: Average of duration and additional Logic

#6 2018-05-15 01:04

Re: Feature suggestions: Average of duration and additional Logic

#7 2018-05-15 07:35

Re: Feature suggestions: Average of duration and additional Logic

Board footer