Duet Wifi Webserver not responding

Nuvag · 2 Feb 2018, 18:43

Updated Webserver to 1.21.RC1 and immediatly having an instable connection, debug log:

WiFi: fGLUE: fragmented pbuf (932!=1514)!
WiFi: GLUE: fragmented pbuf (932!=1514)!
WiFi: GLUE: fragmented pbuf (932!=1514)!
WiFi: GLUE: fragmented pbuf (932!=1514)!
[...]

When reloading the page:

Unexpected state change on socket 0
Can't send anymore
New conn on socket 0 for local port 80
HTTP connection accepted
Found responder
New conn on socket 5 for local port 80
HTTP connection accepted
Found responder
Received 330 bytes
Sending reply, file = yes
HTTP req, command words { GET / HTTP/1.1 }, parameters { }
Received 339 bytes
Sending reply, file = yes
HTTP req, command words { GET /css/dwc.css HTTP/1.1 }, parameters { }
New conn on socket 0 for local port 80
HTTP connection accepted
Found responder
Received 357 bytes
Sending reply, file = yes
HTTP req, command words { GET /js/dwc.js HTTP/1.1 }, parameters { }
WiFi: GLUE: fragmented pbuf (788!=1514)!
[...]

Still having the same issue with http://192.168.178.15/js/dwc.js

When the connection is stable there are no messages with 'WiFi: GLUE: fragmented pbuf <…>', so this might be the cause of the problem.

Edit:
Additionally there is a problem with the html in line 1708:

should be

Nuvag · 3 Feb 2018, 18:09

Webinterface was online 6 hours without a problem then failed again.
I used Wireshark to capture the network traffic, maybe this can help fpr further analysis.

This happened in the moment the connection was lost:
https://www.dropbox.com/sh/r9zkw9id482i9z0/AAA-x0NTtIISPxijZnqTiZj0a/Wireshark_DuetWifi.PNG?dl=0
This was from Microsoft Edge Browser. So my PC sent an reset and from then on all requests to the duet wifi return a reset. This continues until the ajax retry limit is reached (i set it to 50 retries).
I found other occurences earlier where my PC closed the connection but a new one could be established without problems.

Then tried to open the webinterface in a different browser (Firefox) and got:
https://www.dropbox.com/sh/r9zkw9id482i9z0/AADPnonPL1ApvNZzF8qBnV-Va/Wireshark_DuetWifi_2.PNG?dl=0

I suspect for some reason all available sockets from the wifi module in the duet to the internal modules get blocked and won't be unblocked until restarting the wifi. This would explain why the wifi connection works in general (can ping and open tcp connection) but it fails when trying to get the status report.

dc42 · 3 Feb 2018, 21:32

Thanks, I suspected that these problems might be caused by a lack of sockets. Which version of DuetWiFiServer are you using?

Nuvag · 4 Feb 2018, 08:25

Firmware Name:
RepRapFirmware for Duet WiFi
Firmware Electronics:
Duet WiFi 1.0
Firmware Version:
1.20 (2017-12-23)
WiFi Server Version:
1.21RC1
Web Interface Version:
1.20

parovoZZ · 4 Feb 2018, 12:30

Hi!

Is it possible to make a program reset of the Wi-Fi module on an external button? For example, execute a script:

M552 S-1;

M552 S1;

?

dc42 · 4 Feb 2018, 15:35

Yes, if you set up a trigger using M581.

T3P3Tony · 4 Feb 2018, 15:58

if you do that I think you will need a pause in between the M552 S-1 and M552 S1 commands (I tried to make a macro for this but the S1 was executed too soon after the disable (while the wifi was still changing state) so it did not work.

parovoZZ · 4 Feb 2018, 19:19

And then what to do?
Now I connect a USB cable during printing…...brrrrrr))))

dc42 · 4 Feb 2018, 21:10

You can insert a pause between the two M552 commands using G4.

parovoZZ · 5 Feb 2018, 06:26

G4 is bad. Stops printing ….
Is there a NOP command?

dc42 · 5 Feb 2018, 10:46

Good point, if a print is running then G4 will wait for everything to stop. Try this:

M552 S0 ; disconnect and shutdown wifi services
M587 ; list access points, with luck this will wait for the shutdown to complete
M552 S-1
M552 S1

Danal · 6 Feb 2018, 02:12

Another data point:

I lost connection to a DuetWifi during a long print. About six hours in. M582 (via PanelDue) stop/start did not fix. M552 stop/start and connectivity was fine.

Firmware Name: RepRapFirmware for Duet WiFi
Firmware Electronics: Duet WiFi 1.0
Firmware Version: 1.20 (2017-12-23)
WiFi Server Version: 1.20
Web Interface Version: 1.20

Wireless router is a Google WiFi mesh. It is about 15 feet from the printer. It is also true that my Desktop PC is cabled to that same Google node that serves the printer, and the wireless mesh from that node is the only connectivity available to my PC. Meaning that I'd know it instantly if the router mesh node itself dropped.

As mentioned, just another data point. No explicit reply needed.

Nuvag · 11 Feb 2018, 09:05

As im a software developer myself i know such bugs can be hard to track down.
Some minor information which may or may not help:

wifi may fail even when the printer is not printing (turn printer on, connect webinterface, wait some time, connection lost)
wifi may fail within a few seconds after restart. Looks realy random, sometimes hours, and today it failed after <10 seconds

Its a really annoying problem, probably will finally add my PanelDue which i bought with the printer but never bothered to mount it as the wifi interface did not leave anything to be desired.

madeinta1wan · 23 Feb 2018, 13:59

I too have lost connection. I've got long prints running (4-5 days) and am currently 'printing blind'

Is connecting through a USB a bad idea while a print is running? Is it possible to reset the wifi connection from the duet board to the router while a print is running?

Thanks,

Nate

dc42 · 23 Feb 2018, 14:05

Connecting USB while the print is running is OK if you are careful, see https://duet3d.dozuki.com/Wiki/USB_ground_loops. Yes you can reset the network interface during a print.

Have you checked the points at https://duet3d.dozuki.com/Wiki/WiFi_disconnections_and_AJAX_timeout_errors ?

EasyTarget · 23 Feb 2018, 14:22

You should be fine with a laptop and a USB cable; You'd almost certainly be OK even if you put it on charge (laptops typically stay isolated behind their powerbrick).
Once connected You can send [c]M552 S0[/c] then [c]M552 S1[/c] to turn off/on the wifi and it /should/ reconnect to the AP, Use [c]M122[/c] to see how that works out.
I have a RasPI with a USB link to my Duet, and have had to do this several times when the Webcontrol stopped responding during a print, and it was successful each time.

madeinta1wan · 23 Feb 2018, 15:00

That's great, thanks dc42 and EasyTarget . All sorted.

RasPi as terminal server?

Only 121hrs to go…

Nuvag · 27 Feb 2018, 07:47

Is there an fix in the works?
I'm starting to consider downgrading as in earlier versions the wifi was unstable too, but a simple refresh of the webpage reconnected the webinterface compared to restarting the wifi module via USB now.

Yesterday i noticed i can no longer view my Gcode Files in the webinterface as this causes an Request timeout and disconnects the webinterface (but can be reconnected without restarting the wifi)
In earlier versions this did not happen but the list of files was not always complete.

dc42 · 27 Feb 2018, 08:49

@Nuvag:

Is there an fix in the works?
Yesterday i noticed i can no longer view my Gcode Files in the webinterface as this causes an Request timeout and disconnects the webinterface (but can be reconnected without restarting the wifi)
In earlier versions this did not happen but the list of files was not always complete.

There is a separate thread on this. That problem has been seen recently when you have very large GCode files on the SD card. The workaround is:

Format the SD card to use 64K clusters, which is the largest cluster size possible
If the SD card is heavily fragmented, defragment it, or get a new card
Use a fast SD card. I get excellent results using a Sandisk 16Gb Class 4 card formatted to 64K clusters.

deadwood83 · 3 Feb 2018, 02:46

Another user with frequent timeouts. Firmware 1.2.0 WebServer 1.2.1RC4 and WebServer 1.2.0
WAP: Avaya WAP9122
Router: Mikrotik CCR1009
Switch: Avaya VSP4850GTX+

No real way to connect the PC to the device for debugging sadly. Not without a 150ft USB.

7:36:58 PM
Disconnected.
7:33:07 PM
Connection established!
7:28:48 PM
Disconnected.
7:23:46 PM
Connection established!
7:21:48 PM
Disconnected.
7:19:39 PM
Connection established!
7:19:39 PM
Page Load complete!

Other devices (ecobee, phones, etc) are rock solid for months on end.

ETA: M122 Output
WiFi module is connected to access point
Failed messages: pending 0, notready 0, noresp 0
WiFi firmware version 1.21RC3(28b1)
WiFi MAC address 2c:3a:e8:0a:fe:52
WiFi Vcc 3.39, reset reason Turned on by main processor
WiFi flash size 4194304, free heap 13000
WiFi IP address 10.10.30.86
WiFi signal strength -46dBm, reconnections 0, sleep mode modem
HTTP sessions: 1 of 8
Socket states: 2 0 0 0 0 0 0 0
Responder states: HTTP(1) HTTP(0) HTTP(0) HTTP(0) FTP(0) Telnet(0)

DIsconnections are equally bad when printing and when idle. The disconnects above are when idle.