Wednesday, April 24, 2013

CDE: desktop messaging system could not be started & VNC grey screen HP-UX

             
              While running vnc you will get an error "CDE: desktop messaging system could not be started" when you click ok vnc will only show grey screen. [host : uxmemoir01.unixmemoirs.com]

As per different google search results  DNS issue is main culprit. we should have nsswitch.conf like below.
Then only CDE will work properly.

hosts:          files [NOTFOUND=continue] dns

as part of initial trouble shooting in  nsswitch.conf  above line was commented out. I uncommented it out and then  CDE  in vnc client is showing blue screen. hmm.. some progress.

Then I checked /etc/hosts and found that uxmemoir01.unixmemoirs.com is  having proper IP, 192.168.1.1

Then I checked name  servers mentioned in /etc/resolv.conf and found that these name servers are giving another IP for host uxmemoir01.unixmemoirs.com- 192.168.2.19

Issue is that CDE is checking  /etc/hosts and nameservers for hostname look up. It gots wrong IP from name server. So CDE hangs at blue screen.

moral : CDE is extremely sensitive to DNS

Work around :

Comment out the name server entries before starting VNC server and uncomment after starting vnc.

Different places to look for CDE issues.

When CDE is initialized, it must be able to resolve its hostname, ensure that it has access to all hosts, and resolve those hosts defined for the system.
  • Who is affected? If all users are not affected, first check the following list.
    • Filespace in /home and /var
    • Permissions on /dev/null (666)
    • Permissions on /dev/lft0 (666)
    • Permissions on /dev/tty (666)
    • Permissions on /dev/console (622)
    • Permissions on /etc/resolv.conf (644)
    • Permissions on /usr/dt/bin files
  • Check the bold commands and filenames on the left side of the list below and compare the output on your machine with the corresponding information on the right.
    • df    -     Check for full filesystems.
    • hostname    -    Should return hostname of machine.
    • host <hostname>    -    Should return fully qualified machine name and IP address.
    • host <IP address>    -    Should return fully qualified machine name and IP address.
    • uname -n    -    Should return hostname.
    • ping <hostname>    -    Should be pingable, check for packet loss.
    • ping <IP address>    -    Should be pingable, check for packet loss.
    • host loopback    -    Should return localhost is 127.0.0.1.
    • host 127.0.0.1    -    Should return localhost is 127.0.0.1.
    • ping loopback    -    Should be able to ping - check for packet loss.
    • ping 127.0.0.1    -    Should be able to ping - check for packet loss.
    • /etc/hosts    -    Check for duplicate/erroneous entries and permissions.
    • /etc/resolv.conf    -    Nameserver should be pingable.
    • /etc/netsvc.conf - This file may or may not be present, check for typing errors. In this file, nsorder should be hosts=local,bind
    • /etc/X0.hosts - This file may or may not be present, if it is move it to X0.hosts.old.
    • netstat -in - Check to ensure no interfaces have an "*" (asterisk).
    • lssrc -s inetd - Should be active.
    • lssrc -s portmap - Should be active.
    • lssrc -g tcpip - Check to see what TCP/IP deamons are running.
  • If checking the preceding commands or files does not resolve the problem, look for errors in the following files:
    • /var/dt/Xerrors
    • $HOME/.dt/startlog
    • $HOME/.dt/errorlog
  • If no errors exist in the preceding files, your system may have an environmental setting that is prohibiting the login. Check the following files:
    • /etc/environment
    • /etc/profile
    • $HOME/.dtprofile
    • $HOME/.kshrc
    • $HOME/.login
    • $HOME/.cshrc
    • echo $LIBPATH - Should report nothing.
    • echo $MANPATH - Should report nothing.
Source: http://bio.gsi.de/DOCS/www.coredumps.de/doc/ibm/aix/aix_hints_and_tips/6820474.htm