We now get "SystemException: OBJECT_NOT_EXIST (Session/lookup - @x.x.x.x)"
messages on a fairly frequent basis. This is not accompanied by the
'CosNaming::NamingContext::NotFound' message that you would usually get if the
components were not installed properly.
The error does not occur on any particular component, but on a wide range of
components, and often when the user has been working quite happily for some
time. When the user logs back in and repeats the failed operation, it will work
fine.
The components at fault are simple pooled, stateless PB NVOs.
If there is a problem in the CORBA object lookup, is there any way of tracing
what it's doing?
Could this be related to the versions of the PB DLLs being used?
Our client app is distributed with libjcc, libsybheap, pbdwe80, pbrtc80 and
pbvm80 from PB8.0.4. We also upgrade the servers with various PB runtime files,
since 4.2.2 comes with a lower build of PB. These updated runtime files are
installed in Sybase\EAServer\Powerbuilder. This means that the servers have
different versions of libjcc.dll etc in this Powerbuilder directory from the
versions in EAServer\DLL. That is correct, according to the release note for
build 10575, but we have had problems in the past with mixing dll versions, and
have received mixed messages about what to do with the copy of libjcc in
EAServer\dll.
In another strange twist, the error only occurs on 3 machines that are
clustered, but not on a fourth machine in the cluster. Other, non-clustered,
servers have no problems. And if we delete the cluster in JagManager, the
machines are still affected.
Any help would be appreciated.