Created attachment 501624 [details]
RPM referenced in the test case (however, found that test would fail even if Nessus plugins were not installed - reported in step 15)

Description of problem:

The nessusd service seems to leave lingering nessusd processes after a client connection exits.

 * Start nessusd service
 * Bring up a nessus client and log in
 * Terminate ness client
 * Observe nessusd process spawned for client still running and consuming 100% CPU


Version-Release number of selected component (if applicable):

[root@nessus ~]# rpm -qa | grep nessus
nessus-core-2.2.11-6.fc15.i686
nessus-libraries-2.2.11-4.fc15.i686
nessus-client-2.2.11-6.fc15.i686
nessus-server-2.2.11-6.fc15.i686
nessus-gui-2.2.11-6.fc15.i686


How reproducible:

Very

Steps to Reproduce:

1. Install Fedora 15 32 bit from Live ISO

2. Update the Fedora 15 32 bit packages

  yum update

3. Install nessus packages via:

  yum install nessus-server nessus-client nessus-gui

4. Added nessus user by running:

  nessus-adduser
  ...
  Login             : pkb
  Password          : ***********
  DN                :
  Rules             :
  default accept


5. Set the nessus certificate (just pressed enter to take the defaults):

  nessus-mkcert

6. Installed GPL version of Nessus plugins (these do not appear to be available in Fed\
ora):

  rpm -ivh /tmp/nessus-plugins-gpl-2.2.11-4.nst15.i686.rpm

7. Updated the /etc/nessus/nessusd.conf file with the following line (for plugins from\
 above):

  plugins_folder = /usr/lib/nessus/plugins

8. Started the nessus service

  [root@nessus ~]# systemctl start nessusd.service
  [root@nessus ~]# systemctl status nessusd.service
  nessusd.service - LSB: start and stop the nessus daemon
                  Loaded: loaded (/etc/rc.d/init.d/nessusd)
                    Active: active (running) since Sun, 29 May 2011 11:02:01 -0400; 2s\
 ago
                     Process: 8900 ExecStop=/etc/rc.d/init.d/nessusd stop (code=exited\
, status=0/SUCCESS)
                      Process: 8913 ExecStart=/etc/rc.d/init.d/nessusd start (code=exi\
ted, status=0/SUCCESS)
                      Main PID: 8918 (nessusd)
                        CGroup: name=systemd:/system/nessusd.service
                                  └ 8918 nessusd: waiting for incoming connections

9. Looked at number of nessusd processes running:

  [root@nessus ~]# ps -fC nessusd
  UID        PID  PPID  C STIME TTY          TIME CMD
  root      8918     1  0 10:47 ?        00:00:00 nessusd: waiting for incoming co

10. Brought up the nessus GUI using GNOME menu entry and logged in, then looked at number \
of nessusd processes:

  [root@nessus ~]# ps -fC nessusd
  UID        PID  PPID  C STIME TTY          TIME CMD
  root      8918     1  0 10:47 ?        00:00:00 nessusd: waiting for incomin
  root      8994  8918  4 10:48 ?        00:00:00 nessusd: serving 127.0.0.1

11. Quit the nessus GUI client and looked at the number of processes:

[root@nessus ~]# ps -fC nessusd
UID        PID  PPID  C STIME TTY          TIME CMD
root      8918     1  0 10:47 ?        00:00:00 nessusd: waiting for incomin
root      8994  8918 11 10:48 ?        00:00:08 nessusd: serving 127.0.0.1

12. Noticed that process 8994 was still running. Running top showed that this process was \
consuming 100% CPU (pegged).

13. Tried stopping the nessusd service and then checking on what nessusd processes were ru\
nning. Still shows process 8994 running (though the main process waiting for connections t\
erminated normally).

  [root@nessus ~]# service nessusd stop; sleep 10; ps -fC nessusd
  Stopping nessusd (via systemctl):                          [  OK  ]
  UID        PID  PPID  C STIME TTY          TIME CMD
  root      8994     1 56 10:48 ?        00:02:13 nessusd: serving 127.0.0.1

14. So, for whatever reason, these child processes spawned by nessusd do not seem to termi\
nating nicely and leave the system in a bad state.

15. I uninstalled all of the "nessus" packages and then repeated the above steps, but skip\
ping steps 6 and 7 and just setting up Nessus without any plugins (everything at it's defa\
ult state "as shipped in Fedora"). I encountered the same issue. The following command was\
 used to remove Nessus:

  yum remove $(rpm -qa | grep ^nessus-) && rm -fr /etc/nessus/


  
Actual results:

Observe the nessusd process which was forked to handle the client still running:

  UID        PID  PPID  C STIME TTY          TIME CMD
  root      8994     1 56 10:48 ?        00:02:13 nessusd: serving 127.0.0.1



Expected results:

Expected the nessusd child process forked to be terminated once the client closed the connection. Also, expected all nessusd processes to shut down (be stopped) when stopping the service.

Additional info:

As the Nessus packages are very old, it may be that they are being phased out of inclusion in the Fedora repositories (possibly being replaced by the openvas packages). If this is the case, maybe the simplest solution is just to remove the nessus packages from the repositories.