Bugzilla – Bug 3412
Staging delegation host address needs to use logical host name
Last modified: 2005-05-31 11:47:56
You need to
before you can comment on or make changes to this bug.
I am able to run a simple fork or LSF type jobs with globusrun-ws, but when
staging/streaming the credential delegation service errors due to the target
address not being what was expected in the stagingDelegationFactoryEndpoint.
Cleanup is having this problem with the host address . Apparently not using
the logicalHost of the container . My hostname and container logicalHost were
not the same. Hostname was "f3n2e.....", later I set logicalHost in the
globus_wsrf_core/server-config.wsdd after I installed GT4 to the name which
a remote client will use. Here is the error:
Cleaning up any delegated credentials...Done.
globusrun-ws: Job failed: Staging error for RSL element fileCleanUp.
; nested exception is:
org.globus.common.ChainedIOException: Authentication failed [Caused
by: org.ietf.jgss.GSSException, major code: 15, minor code: 25
major string: Operation unauthorized
minor string: Authorization failed. Expected
"/CN=host/f3n2e.fnmoc.navy.mil" target but received "/C=US/O=U.S.
I expected to get the same hostname (or ip address ) used in the ws address
for delegationFactoryEndpoint and stagingDelegationFactoryEndpoint , but
instead I get the container logicalHost in the first, and the original
hostname in the
Should the stagingDelegationFacrotyEndpiont be using the logicalHost that I
have specified in
globus_wsrf_core/server-config.wsdd ? What means can be used to reconfigure
Here is the response shown by "globusrun-ws -dbg" showing the different
The endpoint is constructed in the GRAM code AFAIK. I am going to wait for GRAM
folks to comment on this before investigating of the delegation service code.
In etc/gram-service/jndi-config.xml, the pararmeter value for stagingHost
was changed to the same value used for the logicalHost, now it works.
I'm marking this as a duplicate of bug 3132 since this should not be an issue
once that is fixed and
because this clearly has a workaround.
*** This bug has been marked as a duplicate of 3132 ***