Timeout waiting for packet

[BIRD]

Hi
Want to ask in the burn in test network option :
Bad packet generates error
Bad packet ratio generates error


We test default 4000ms run network 100% --> result FAIL (Timeout waiting for packet)
We test Ratio 2% run network 100% -->result PASS (no error)

What's the difference between the above two settings?

Thanks.

[Richard (PassMark)]

Bad packet generates error will report an error for every error. Bad packet ratio generates error only if a certain ratio threshold is reached, e.g. 2% is 1 error in 50 packets. Per the help file documentation, Help->Help File, "Selecting Test Preferences->Standard Network Test and Bluetooth":

Bad packet generates error / Bad packet ratio generates error / Ratio (Error handling)

There are two options for handling errors that occur during the execution of the network test. BurnInTest can either log an error for every bad (or missing) packet or can be set to only log an error when the number of errors exceed a threshold ratio set by the user. This threshold ratio is expressed as the percentage of bad packets compared to the overall number of packets attempted to be sent.

Timeout

This value determines how long BurnInTest will wait for a data packet to be sent or received before an error is reported. The value is measured in Milliseconds. 2000ms (2 seconds) is the default value.

[BIRD]

I would like to set 4000ms test result FAIL; set 2% test result PASS.
Which judgment do we base on?
The current actual device test and signal quality test results are PASS.
We just want to confirm that the difference of 4000ms / 2% test results will make us confused.

[David (PassMark)]

Sorry I don't really understand the question.

[BIRD]

We test default Bad packet generates error 4000ms run network 100% --> result FAIL (Timeout waiting for packet)
We test Bad packet ratio generates error 2% run network 100% --> result PASS (no error)

But the performance of my network device is good.
How to explain the above two different results ?

[David (PassMark)]

It is normal to have some errors on a network connection.
TCP/IP itself has flow control algorithms that depend on errors to work.
There are also checksums, packet numbering & retransmissions to ensure data isn't lost.
So a small number of errors isn't of any great concern. What's a reasonable number of errors really depends on the setup, e.g. Wifi at long range would be expected to generate many more errors than Ethernet.

[BIRD]

Hi David
I see, thank you for your help.