Archived

This forum has been archived. Please start a new discussion on GitHub.

IceGrid Registry service suddenly stops working after reboot/restart

Hello,

we are trying to set up a IceGrid Registry with several distributed services, which should be running in the same replica group to perform a load balancing across all servers. The IceGrid Registry service and our service nodes are running as installed Windows services. Our registry servers (as QS-AWBAWPV2 in our screenshots) have “Windows 2012 R2” installed, on our clients there are some Windows 8 (as KERNELDIH) and some Windows 8.1.

After the first installation everything works on our testing machines like a charm, even after we do a reboot or a service restart. But unfortunately on some servers we discover a weird error after we reboot or just restart the IceGrid Registry service. The registry service won’t start and returns a 1 as error code. As we look up in the windows event log, we discovered this two messages
• can't add adapter `KERNELDIH-Infrastructure-Collector.Singleton.AWIceBoxAdapter' to unknown replica group `collectorServiceRepGroup'
• service caught unhandled std::exception:
C:\Users\jose\Desktop\RC3\build-x86-VC100-release\Ice-3.5.1-src\cpp\include\IceUtil/Handle.h:46: IceUtil::NullHandleException

The only solution we figured out for this problem, is to db_recover the registry database to a date before adding any client services. Then the registry starts again and we could add all client services again until we restart again.
Strange thing is, all of the servers are installed with the same tools and with the same configurations, but some work and on others it will crash every time after reboot/restart.

Now we are out of ideas, what to do.
Are we doing something completely wrong? But then, why is it working in some cases?
Or are these kind of errors known and have an easy solution?

I’m attached a corrupt registry database, screenshots from the event log errors, a IceGridAdmin screen and our configuration of the registry and two service nodes.

Best regards
Dirk Holzapfel

Comments

  • benoit
    benoit Rennes, France
    Hi,

    The source patch for this issue can be download at this thread.

    Cheers,
    Benoit.
  • Hi,

    thanks for providing the patch for this issue. It resolved our problem entirely.

    Thanks and Best regards
    Dirk Holzapfel