bin/51586: rsh/rshd connect problem (select: protocol failure in circuit setup)

Tue Apr 29 07:40:33 PDT 2003

>Number:         51586
>Category:       bin
>Synopsis:       rsh/rshd connect problem (select: protocol failure in circuit setup)
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Apr 29 07:40:18 PDT 2003
>Closed-Date:
>Last-Modified:
>Originator:     Lars Köller
>Release:        FreeBSD 4.8-RELEASE i386
>Organization:
Computing Center, University of Bielefeld, Germany
>Environment:
System: FreeBSD vscan2.hrz.uni-bielefeld.de 4.8-RELEASE FreeBSD 4.8-RELEASE #0: Fri Apr 11 12:16:31 CEST 2003 root at vscan2.hrz.uni-bielefeld.de:/usr/src/sys/compile/VSCAN i386

>Description:

- The setup is: machine1 (FreeBSD or Solaris 8) connects to maschine2 (FreeBSD)
  via rsh.

  On machine1:
  while true; do 
      /usr/bin/rsh maschine2 -l vscan pwd; ret=$?
      if [ "$ret" != "0" ]; then 
          echo "Return Code: $ret
	  break
      fi
  done

- On the comman line one sees:

  /usr/home/vscan
  ..........    (repeats several hundred times)
  ..........    
  /usr/home/vscan (hangs, and after 120 seconds)  
  select: protocol failure in circuit setup

  The rshd logs:  Apr 29 16:04:17 vscan2 rshd[2308]: connect second port 623: Operation timed out

- inetd is running as:

  /usr/sbin/inetd -w -W -l -C 256 -c 64 -s 64 -R 0

  I tried to start it without -w -W, but this didn't change anything!

- inetd.conf for rsh:

  shell   stream  tcp     nowait/0/0/0 root       /usr/libexec/rshd       rshd

- To avoid a TIME_WAIT blocking of the ip:port quadruple for the connection,
  which was our first guess, I set on both machines (FreeBSD)

  sysctl -w net.inet.tcp.msl=300

  After this you cant see ant TIME_WAIT backlog with netstat -a

  (I need a high rsh connection rate, cause it should be used in a farm setup
   for mail virus scanning).

- /etc/hosts.allow

  rshd: machine1        : allow

- ipfw:

73869 : ipfw -ta show 
00100  1450976  204275072 Tue Apr 29 16:23:59 2003 allow ip from any to any via lo0
00200        0          0                          deny ip from any to 127.0.0.0/8
00300        0          0                          deny ip from 127.0.0.0/8 to any
65000 16410061 6975271579 Tue Apr 29 16:23:59 2003 allow ip from any to any
65535        0          0                          allow ip from any to any

- When I do the same on machine2 with 

  .....
  /usr/bin/rsh localhost -l vscan pwd ......
  .....

  instead of machine2 (see while loop above) all seems to work properly!
  It runs and runs and runs

>How-To-Repeat:

  See the above setup/description

>Fix:

  Don't know, but I need one very fast.

Any ideas???

-- 
E-Mail: Lars.Koeller at Uni-Bielefeld.DE        \  Lars Köller
            lkoeller at FreeBSD.org              \  CC University of
PGP: http://www.uk.pgp.net/pgpnet/wwwkeys.html \  Bielefeld, Germany 
     Key-ID: A430D499                           \  Tel: +49 521 106 4964
----------- FreeBSD, what else? ---- http://www.freebsd.org -------------
>Release-Note:
>Audit-Trail:
>Unformatted: