William Boehm
2013-08-12 18:28:37 UTC
Hi,
We are running OmniOrb 4.1.4 on Solaris and are having some random connectivity issues after our nightly bounce. Our system is makeup of a number of C++ modules using OmniOrb and a number of JAVA 1.6 modules using the JAVA implementation of CORBA and we are using the JAVA transient naming service.
As part of our startup routine we have a module that connects to an ORACLE database and loads up information into memory when this modules has completely stated and registered itself with the naming service the next module will try to start up and communicate to the first module to load information specific to that module in its cache. The modules all startup ok and appear to be up and running. Our system is such that we do the nightly bounce and there may not be any traffic flowing thru the system for hours depending on the location of the client. What we are seeing is that maybe 1% or less of the time (we have gone weeks without seeing it and then have it happen two days in a row) when the next morning comes we receive a request for JAVA to one of our modules (we have seen this happen on both C++ and JAVA) but it is unable to communicate with the module and the only way to correct it is to bounce the module and depending on which one of our modules has the problem it might require a bounce of the whole system.
What I am wondering is has anyone else ever seen behavior like this and if so how did you go about correcting it? If not does anyone have any ideas as to how we can get to the bottom of this as we currently have no way of knowing if a system has this problem until we either receive a call from someone reporting the issue or when we tail the logs and don't see any traffic going thru the system.
Thanks,
Bill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.omniorb-support.com/pipermail/omniorb-list/attachments/20130812/0ada5c57/attachment.html>
We are running OmniOrb 4.1.4 on Solaris and are having some random connectivity issues after our nightly bounce. Our system is makeup of a number of C++ modules using OmniOrb and a number of JAVA 1.6 modules using the JAVA implementation of CORBA and we are using the JAVA transient naming service.
As part of our startup routine we have a module that connects to an ORACLE database and loads up information into memory when this modules has completely stated and registered itself with the naming service the next module will try to start up and communicate to the first module to load information specific to that module in its cache. The modules all startup ok and appear to be up and running. Our system is such that we do the nightly bounce and there may not be any traffic flowing thru the system for hours depending on the location of the client. What we are seeing is that maybe 1% or less of the time (we have gone weeks without seeing it and then have it happen two days in a row) when the next morning comes we receive a request for JAVA to one of our modules (we have seen this happen on both C++ and JAVA) but it is unable to communicate with the module and the only way to correct it is to bounce the module and depending on which one of our modules has the problem it might require a bounce of the whole system.
What I am wondering is has anyone else ever seen behavior like this and if so how did you go about correcting it? If not does anyone have any ideas as to how we can get to the bottom of this as we currently have no way of knowing if a system has this problem until we either receive a call from someone reporting the issue or when we tail the logs and don't see any traffic going thru the system.
Thanks,
Bill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.omniorb-support.com/pipermail/omniorb-list/attachments/20130812/0ada5c57/attachment.html>