Our extension, when running in InDesign CS 6, we are getting intermittent network failures that [usually] show up as timeouts.
* Our server is running at AWS and is publicly available.
* The error shows up on multiple machines, but all physically located in the same office; so far, we have not had reports from other locations.
* So far, the problem has only been reported in CS 6; our extension also runs in CC, but the problem has not been reported there.
* Running the same request from a browser works and is quick. (tested with Safari & Chrome).
* The error is extremely erratic: it will happen repeatedly on one machine for an hour, then disappear for the rest of the day or could happen for one call, the next call will succeed and the third call will fail, even though all calls are sent in rapid succession.
* Sometimes it fail with a timeout, other times there will be no success or failure, it just never returns; on one occasion, we waited for 15 minutes, but our listeners were never called.
* The behavior does not appear to have any connection to the load on the server.
Some Errors Details:
|Server.Error.Request||versionCheck||HTTP request error||Error: [IOErrorEvent type="ioError" bubbles=false cancelable=false eventPhase=2 text="Error #2032" errorID=2032]. URL: home/info/apiVersion.txt|
|Client.Error.RequestTimeout||fetchPreferences||Request timed out||The request timeout for the sent message was reached without receiving a response from the server.||I|
|Client.Error.RequestTimeout||savePreferences||Request timed out||The request timeout for the sent message was reached without receiving a response from the server.|
* We use HTTPService to send the request
* [Nearly] all of our calls go thru a common class which creates the HTTPService, adds parameters, add listeners, etc.
Things we have tried:
* Upgrade everything: FlashPlayer, Java, InDesign CS 6
* Check server log: as far as we can tell, the call never reaches the server; we are looking at both our own log and localhost_access_log (our server runs in tomcat).
* Network connectivity: as far as we can tell, there are no network issues on the machine. During our investigation, when the error occurs in the plugin, we immediately open the web interface, on the same server, and perform the same action. The web interface always works and responds quickly, so that machine can connect to the server.
* Some times it works: if we had a badly constructed request or parameters, or we had general network access problems, the errors would be consisted, but the extension works fine for all operations, sometimes and on some machines.
One additional note: our initial data gather has been clouded by real time outs (call reached the server but it took too long to create and send the respond); we have addressed these problems but are still seeing the mystery errors (which appear to never reach the server).
Help: I'm looking for any advice or guidance on what to investigate. It seems like the problem is in network layers on the machine, but it is difficult to isolate.
Any suggestions from coding changes to plugin building to machine maintenance would be welcome.