Details
-
Bug
-
Resolution: Invalid
-
Not Evaluated
-
None
-
4.7.3
-
None
-
Windows
Description
Execution context:
- Running as a Windows Service, with the Local System Account
Normal behavior:
- Opening an SSL connection every 5 minutes, exchanging a few bytes (around 5k), closing
The application is correctly working on 5000+ machines. On 6 or 7 machines, our application is working correctly for a variable amount of time until network operations are failing with an "Unknown error". This behavior lasts exactly 24 hours after a connection is finally working, and then it's blocked for again exactly 24 hours. If the application is restarted, everything works well, until it fails again, with the previously described behavior.
Implementation details:
- The application is using synchronously the Network component (waitFor*())
Observations:
- The machine is not having any particular stress
- There is nothing similar on machines (OS, architecture granularity, anti-virus, firewalling rules)
Already investigated without any results:
- DNS flushing
- Windows logs are not talking about anything
- There is no quotas restrictions
- Others machines on the same network are working well
Global details:
- Our application is a monitoring tool, and we are graphing the network usage. I added two attachments files. The first one (success.png) is a normal graph, the second one (fail.png) shows the exactness of the working communications
Attachments
Issue Links
- relates to
-
QTBUG-24451 windows - QAbstractSocket::waitForXXX functions are unreliable
- Open