Hi all, While debugging on "why does not ports used by ganesha cleanup properly", I found out the following issue :
Ganesha got killed before closing main thread and admin thread . I will try to explain what happens from my understanding of the code. When I perform "service nfs-ganesha stop", sigmgr_thread receives a "SIGTERM" message then it will send `admin_shutdown` message to admin_thread and it will wake up. In admin_thread , do_shutdown() api which will close all the threads , remove the export and unregister all the ports. But in the do_shutdown while closing the worker threads , ganesha process got killed without any generating cores. From the logs I am sure that admin_thread , main_thread, one or more worker threads are still in waiting state. It is 90% reproducible for me (only starting and stopping of ganesha service is done) . And also since Clean_RPC() function is called after the woker_shutdown() , ports does not unregister properly. I am working in RHEL6.7 vm and using latest ganesha code. Any thoughts on why ganesha process got killed before closing all its threads? With Regards, Jiffin ------------------------------------------------------------------------------ _______________________________________________ Nfs-ganesha-devel mailing list Nfs-ganesha-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs-ganesha-devel
