Burn-in-Test Network test error "Error sending packet. Check destination address."

[mingtso]

Hi,

I use Burn-in-test v9.1(100 and v9.2(1009) for run-in test.
But I got the fail with network standard test on both BIT version.

My test device has three network interfaces.
I assign these adapters to different network subnet ex: 192.168.3.33, 192.168.4.33, 192.168.5.33.

I found while the first or second interface get lower link speed (100M) the network order is changed in BIT.
So the BIT sends test packet to incorrect subnet.

LOG NOTE: 2023-03-20 22:01:33, Network 3, NIC: 192.168.3.33 (->192.168.5.1)
LOG NOTE: 2023-03-20 22:01:33, Network 3, Test packet start 0.
LOG NOTE: 2023-03-20 22:01:33, Network 3, Send packet(Sample 1 of 10000).
LOG NOTE: 2023-03-20 22:01:33, Network 3, Send Socket error: 10051
LOG NOTE: 2023-03-20 22:01:33, Network 3, Update statistics.
LOG NOTE: 2023-03-20 22:01:33, Network 2, NIC: 192.168.5.33 (->192.168.4.1)
LOG NOTE: 2023-03-20 22:01:33, Network 2, Test packet start 0.
LOG NOTE: 2023-03-20 22:01:33, Network 1, NIC: 192.168.4.33 (->192.168.3.1)


I also add ping tests before BIT and after BIT test.
The ping test is OK. So network is working fine in OS environment.

I do this test around 500 power cycles. The failure rate is about 3/500.


Could you give some advice how to avoid the test error?

Thanks,

Mingtso

[David (PassMark)]

We no longer support V9. Can you try V10?

Also I think we need some more details of the network environment, and also more details of what you mean by this, "I found while the first or second interface get lower link speed (100M) the network order is changed in BIT"

[mingtso]

Hi David,

My test network topology concept is shown as below. There are three sub networks. And all network adapters are all gigabit ethernet adapter.


All ethernet adapters of DUT are all configured as auto-neg.
In genereal, the network should link as gigabit(1000M) while adapter establish the network connection.
But sometimes, adapter could not link as 1000M and it will try to link with lower speed ex: 100M.

In this test, I don't care the network physical link speed is 1000M or 100M.
The DUT can send packets to server and receive the response, I think the network is working.

But I used BIT v9.1(100 for testing. I got the fail log which was attachmeted in previous post.

My test procedures are listed as below.

1. Set up PC IP address as fixed IP address.
2. Configure BIT network preferences as "standard network test".
3. Set standard network test host address. Configuration example likes below picture.
The configuration will be modified depends on real case.

4. Start test for 5 minutes and check test result.
5. Add auto start BIT test batch(execute: bit.exe /r) into Windows startup folder.
6. Power off test DUT.
7. Power on the test DUT.
8. Windwos startup auto launch bit.exe /r to perform the test. Test duration is 5 minutes.
9. Wait for Bit test finish and power off the DUT.

Test repeats step7 to step9 500 times.
The test failure only happened while port1 physical link is 100M.

In my test case.
My DUT has three adapters EC1000(onboard), EC1000 #2((onboard) and i226(PCIE).
EC1000 is port_1 and IP is 192.168.3.33. (ping to server IP 192.168.3.1)
EC1000#2 is port 2 and IP is 192.168.4.33. (ping to server IP 192.168.4.1)
i226 is port 3 and IP is 192.168.5.33. (ping to server IP 192.168.5.1)

In normal case, all adapters are link as 1000M.
BIT test got PASS result.

In fail case, EC1000 is link as 100M, the others are link as 1000M.
BIT test got FAIL result. ( "Error sending packet. Check destination address.")

Please help to check this issue with BIT 9.2

Regards,
Mingtso

[David (PassMark)]

adapter could not link as 1000M and it will try to link with lower speed ex: 100M

Hard to know from the diagram, but I assume each of these cables is a very short cable run?
If the NIC isn't stable at 1000M then this is generally an indication of some electrical problem.

I see from the log you are using Intel NICs. These had known problems recently. See,
https://www.tomshardware.com/news/in...around-for-now
and
https://www.intel.com/content/www/us...to-2-5gbe.html

So you could try the Intel patch and / or disabling Energy Efficient Ethernet (EEE)

Also, we no longer support V9. Can you try V10?

[mingtso]

Hi David,

I have tested V10 and disabled EEE setting on all ethernet adapter.

I still got the same fail.

I upload two screen shot.

First, both LAN port linked as 1Gbps. There is no error.




But LAN port1 linked as 100mbsp. The issue is occur.

[David (PassMark)]

If the connection is dropped and restarted at a lower speed, then it isn't so surprising that the network connection is interrupted.

[mingtso]

Hi David,


During the test, the network isn't drop or restart at lower linkspeed.

I changed the linkspeed manully before start BIT test.

Why BIT get diffent network interface order to send test packet?

With the lower linkspeed, windows network still work as well. It only gets failure test result in BIT.
What is the propose of BIT standard network test?

Mingtso

[David (PassMark)]

We obviously can't see your network config from here. So hard to guess at the problem. But I would be fairly confident that if there was an error about being unable to send data, it is correct.
Usually meaning that the destination address isn't reachable from the source NIC (so subnet mask setup, destination is wrong, source NIC isn't available, firewall, etc.).