Second call of EJB client causes server stuck in some cases. When this happens administrator is not able to shutdown or reload such a server.
Scenario details:
Elytron is configured as security provider on servers and Elytron client API is used for authentication. Protected stateless beans are used.
EJB Client -> Entry bean on Server1 -> WhoAmI bean on Server2
- Client uses AuthenticationConfiguration with valid credentials for Server1; Client calls the Entry bean twice - the difference is in arguments provided. The arguments are subsequently used for authentication from Server1 to Server2 (using AuthenticationConfiguration):
- first call uses null for both username and password
- second call uses valid credentials for Server2
- Entry bean on Server1 calls WhoAmI bean on Server2 with credentials which was provided as method arguments;
- WhoAmI bean on Server2 just returns caller principal
The Client is executed twice, the second call doesn't finish
- is blocked by
-
JBEAP-11880 EJB server to server fails during transaction commit (Elytron, Remoting)
- Closed
- is cloned by
-
WFLY-9014 Second remote EJB client connection prevents server from shutting down (Elytron, Remoting)
- Closed
- relates to
-
ELY-1272 Missing null checks in auth client lead to NPEs later on
- Resolved