Thanks for the responses. Everything I can see indicates the messages are UTF-8 and thus the messages are legit. That wouldn't explain why sometimes it works and sometimes it fails though. I put the debug on all, but it's logging the bytes of the messages. Looking at that, I can't really tell if the messages are off. One detail I left out is the Observable/Observer pattern I'm using. The Observer implements my DBusInterface and exports itself to the dbus. The Observable is a thread that runs and updates its state every so often. When the Observable updates, the DbusObserver gets the update and emits a signal to the dbus. Is this perhaps a bad design? Could having N Observers accessing the dbus be an issue? (I'm guessing no since the dbus library appears thread-safe. I'm somewhat grasping at straws now.) I went ahead and tried to create a small project to demonstrate the issue (mostly for myself). Again, sometimes it fails, sometimes it doesn't. Sometimes it takes a while to fail. If anyone somehow has the time/desire to dive into it, <a href="https://www.dropbox.com/s/ffbjp24h58s1yg3/dbus-test.tar.gz">https://www.dropbox.com/s/ffbjp24h58s1yg3/dbus-test.tar.gz</a> -- Colby M. White <div class="gmail_quote">On Tue, Sep 4, 2012 at 6:07 PM, Matthew Johnson <<a href="mailto:dbus@matthew.ath.cx" target="_blank">dbus@matthew.ath.cx</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> <div>On Tue Sep 04 11:40, Colby M White wrote: > I am attempting to use the java bindings for dbus and am encountering a > disconnecting problem. When I send a signal, the connection is disconnected > with out given me much of a reason as to why. Part of the issue is it > doesn't happen all the time, just when it feels like it. That's making it > hard to reproduce. </div>Hi, that's really strange - it definitely shouldn't do that. <div> > So my main question: What are the possible reasons that a connection would > decide to disconnect (other then hitting the DBusConnection.disconnect() > method, which I confirmed is never being called in my code)? If I knew what > could cause a disconnection, then it would narrow my search. </div>Generally sending malformed stuff. Aside from strings I really should be stopping you sending anything illegal though. <div> > One unique aspect to my setup is that I'm sending multiple signals, each by > its own thread. As far as I can tell, I've done this in a thread-safe > manner. When the signals are sent, they are sent in a synchronized block. > Here's a quick stream-lined glance at my class (Topic is a basic > DBusInterface that only has a signal): </div>it should already be thread-safe. Sending any message just appends it to a queue, which a background thread pops from and does the actual send. <div> > > Is there something about using this java binding in a threaded environment > that I'm missing? Are there some dbus-thread guidelines/safe practices? I'm > not entirely sure my threaded environment and the disconnections are > related (or at least my tests aren't showing that they are), but the very > nature of threads makes it hard to know. My gut tells me they probably are > related. </div>Nothing in that snippet looks obvious. Do you have a testcase you can reproduce with? <div> > I recompiled my dbus jar with debug enabled and here is what is happening > after I send the signal. As you can see it writes, then disconnects right > after. > > [org.freedesktop.dbus.AbstractConnection$_sender.run()] Got message: > DataChanged(0,7) { Path=>..., Interface=>dbus.topics.Topic, > Member=>DataChanged, Signature=>a{sv} } { { //the contents of my map//} } > [org.freedesktop.dbus.Message.append()] Appending sig: a{sv} data: [{ //the > contents of my map//}] > [org.freedesktop.dbus.MessageWriter.writeMessage()] <= DataChanged(0,7) { > Path=>..., Interface=>dbus.topics.Topic, Member=>DataChanged, > Signature=>a{sv} } { { //the contents of my map//} } > [org.freedesktop.dbus.MessageWriter.writeMessage()] Writing all 102 buffers > simultaneously to Unix Socket > [org.freedesktop.dbus.DBusConnection.disconnect()] Disconnecting > DBusConnection > > Any help on dbus disconnections and/or how to get it to tell me why it is > disconnecting would be helpful. </div>Can you try changing the log levels by adding a debug.conf in the working directory? Try: org.freedesktop.dbus.MessageWriter=VERBOSE or (warning, lots of output) ALL=VERBOSE Matt -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) iQIcBAEBCgAGBQJQRomtAAoJEKVC9ua3o6Nmop0P/jlI1RZDxIFAhwK82IN2o8aN hGHRa5p208bnI9ND3GGt2QNzXA1evCy15CoLIsurMQUlDrqV+uu6OnUelGkkk4da R8enExKwRI94xssdYzb47n8DQBsxxG6Chmu8z9KCUFSlECFGivMwCyt5ptDRvWMv mxN0O6D8LOMsOusm7bA9ceUELYK6B+mtE2UT7yqVvlbnw3U18DtqK7naGx/UwEdL zTpWC00W7gpaBP3f6OPDUcCuOPycvNlv5hNcFgbgSes7rskRCcnk2zOXkX0/MudM cJ4URSVuDvQWGJsdwNs2WW7d0lSHRKvNBfnK289hvr7mpgnIui1v3y+HpX+9epW8 QwAboaKNF/1a3xkxtiClshpZBLDkEsd5oyUOu6H0qOisTEHq+xaXJnXEePa0Y7Vs +5uPzhLKx9Sc2yBwGrVrwA8mTHA+tdC7l5Wwr4JxblR7g0Yv1IQd/xl30SUWd4Cl fo1T17YgC9CnZzrjEaUHPDu8/cD/QNx8HciOI5EbX+slVx0DCYjjA7ZybzqlTMz7 sXdiqE0GhWg17i4biE6/+2iPKn7PXDoUQqPJIrjW0qfN911etebBFkVtugGoeQ8C ltNqWFYpK09jX2XK1QzOrj4EAsHtg0jC8pZEm2G3/rkFlY2xmDX4SSfyDWAOcy1m AJsxwJQbkoMiNmzv1Qyz =ijnA -----END PGP SIGNATURE----- </blockquote></div>