[2.0] lost frame with wireshark

[Raizo62]

Hi

I see the bugs since several days :

Before this post, i have tested with this example (where i have used your capture command) :
I have 2 VMBox connected to an ethernet switch.
After start the VMs, i start wireshark on a link.
I do traffic between these 2 VMs. I see network frames in wireshark.

I stop traffic and i do something else (drink coffee or a configure config file). I don't wait very long time : 1 or 2 minutes. Perhaps 3.

I do new traffic between theses 2 VMs. Traffic is ok but no new network frame in wireshark.

[Raizo62]

2 times that i can't reproduce it.
I hope that i have do an error or the bug is corrected.
I think that you can close it. Except if you can reproduce it (of course ;-))
Sorry for lost time

[Raizo62]

Sorry, i lost again frame. I was believing (hoping) it was because i was using a realtime kernel.
But with classic kernel, wireshark loses again frames after somes times.

(I have even removed this message because i could not again reproduce it after reload project. After test "error with cloud", i can see it, but i don't know if it is linked)

[julien-duponchelle]

@grossmj you have the same issue ?

[grossmj]

I've tested and could not reproduce it.

Connected 2 VMs to a Ethernet switch. Started the VMs, captured on one of the link, stopped the capture, waited 5 minutes, started a new capture and I could still see traffic flowing between the 2 VMs.

[grossmj]

@Raizo62 maybe a video would help... or even screenshots.

Thanks!

[Raizo62]

captured on one of the link, stopped the capture, waited 5 minutes, started a new capture

no. it is not that : I don't stop the capture, but i don't do traffic.
If i stop wireshark (even if bug) and restart : that works.

[grossmj]

You mean you don't stop the capture but stop the traffic from inside the VMs, right?

[Raizo62]

I stop the traffic between 2 VMs.
For example, i stop "ping" or services communications. But i don't shutdown interfaces.

[grossmj]

Ok, I managed to reproduce the issue.

I used 2 VirtualBox VMs running Linux. I shutdown the interfaces for 5 minutes (ifconfig eth0 down) while a capture is running. After re-enabling them and pinging between the VMs, communication still works but no traffic is captured anymore.

@noplay is it possible the capture monitoring timeouts in the server?

[julien-duponchelle]

Since it's HTTP stream I suppose it's possible that something timeout due to no activity. You can go to http://localhost:3080/controller view the project and you can get the pcap link for testing if it's a server or client timeout.

…

On Fri, Jan 13, 2017 at 5:08 PM Jeremy Grossmann ***@***.***> wrote: Ok, I managed to reproduce the issue. I used 2 VirtualBox VMs running Linux. I shutdown the interfaces for 5 minutes (ifconfig eth0 down) while a capture is running. After re-enabling them and pinging between the VMs, communication still works but no traffic is captured anymore. @noplay <https://github.com/noplay> is it possible the capture monitoring timeouts in the server? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1763 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAVFXVWmVJ1qiuVhknb2WsByojmGqhtLks5rR6GAgaJpZM4Lb8jH> .

[grossmj]

Same issue with 2 VirtualBox VMs connected back to back. Also, I have tested with IOS routers and I have the same problem, hence I think there is globally a problem with capture.

The controller shows the capture is still active after 5 mins and re-enabling the interfaces inside the VM.

But nothing new shows up in Wireshark.

Looking at the pcap file, it does not grow after pinging between the node.

[julien-duponchelle]

If I curl the pcap from api the pcap is updated. This mean the issue is inside the Qt code