Full information about endpoint is cached on infoprovider node, which shouldn't be used in that way. It's responsibility is only to extract runtime information about job (input parameters + tags). So changing communication data structures between master<->infoprovider will fix that problem.
Full information about endpoint is cached on infoprovider node, which shouldn't be used in that way. It's responsibility is only to extract runtime information about job (input parameters + tags). So changing communication data structures between master<->infoprovider will fix that problem.