On Mon, Dec 14, 2020 at 4:31 AM Pietro <[email protected]> wrote:
> Hi Nick, > Thank you for your reply. > > As you suggested, last week we repeated the experiments starting a new > (unshared) connection using the 4G Provider 1. Unfortunately, same issue > occurred. > > This time we also noticed that the first error message appears after ~5/10 > minutes, then after the first occurrence the issue repeats every 1 or 2 > min. > Same thing happens with shared connections. > > Attached you find a log obtained by merging by timestamp guacd (in brown) > and Guacamole-client (in blue) logs. > > Server_Client_Log_(1).odt > < > http://apache-guacamole-general-user-mailing-list.2363388.n4.nabble.com/file/t1622/Server_Client_Log_%281%29.odt> > > > In the same file you also find WS messages sent and received by the > browser, > along with their timestamps. > > These tests showed that somehow the guacd was not receiving messages from > Tomcat (or upstream components) on time. > This is not necessarily the case. The tunnel that is established for Guacamole is from the browser to Tomcat over HTTP(S) or WS(S), and then Tomcat opens the socket to guacd. The issue you're seeing - user is not responding - may just be because Tomcat isn't receiving data in a timely fashion from the end user (web browser), so it has nothing to forward on. You mentioned in the previous e-mail that everything worked fine on one 4G network and on WiFi, and, assuming that is still true, this would rule out a communications issue between Tomcat and guacd and point back at the 3G/4G network(s) you're having trouble with. > > Trying to isolate the issue, we did the following: > - Switched from Tomcat 9.0.38 (I had wrongly stated it was version 8) to > Tomcat 8.5.61. But same issue. > - Tried Guacamole version 1.3.0, for both guacamole-client and > guacamole-server (guacd), using the Docker images. But same issue. > > As additional note, I can also say that the browser keeps on receiving and > sending messages fast until disconnection. Likewise, image frames are > received until disconnection (with some delay compared to what the other > users see when using shared connections) > > This adds to my suspicion that something is being done on the 3G/4G network at a network level that is impacting the traffic - the fact that the clients seem oblivious to any problem, and that other clients on other networks seem to work fine with the same server really points to something within the network itself. -Nick
