Distributed SMIT 2.2 for AIX: Guide and Reference

DSMIT Troubleshooting

Use the following suggestions if you encounter problems while connecting to a client, listing domain members, or defining a working collective. A Failed prompt in the Command Status field of the output screen indicates that one or more systems reported a failure.

Clean Up After Abnormal Termination

If the front end of DSMIT/crex/srex abnormally terminates, the following may not have been cleared:

Files in /tmp/dsmit that were created during startup and removed during exit.
Contents of the counter file /usr/share/DSMIT/security/dslock.ctr.
/usr/share/DSMIT/security/dslock may exist.

Concurrent DSMIT/crex/srex sessions may have similar side effects and they must be resolved. The following information is intended as a guide for cleaning up processes that abnormally terminated:

Remove the files in /tmp/dsmit that belong to the failing process.
Use the dsmit-dec command to decrement by 1 the value of the counter in the /usr/share/DSMIT/security/dslock.ctr file.
Remove /usr/share/DSMIT/security/dslock lock file.

If You Cannot Connect to a Client

Many of the problems associated with Distributed System Management Interface Tool (DSMIT) arise when you cannot connect to one or all of the clients. Consult the following list if you cannot connect to the DSMIT clients:

Check to see if the network is down. From the command line, issue the ping command to the client to which you cannot connect:
```
ping ClientName
```
If the client responds, then the network is active. If the client does not respond, follow the problem determination steps defined by TCP/IP.
Ensure that the inetd daemon is active on both the server and client. To ensure that the inetd daemon is active, check the processes for the server and client, or issue the following command from the command line:
```
rsh HostName ps -ef | grep inetd
```
If the inetd daemon does not exist or the rsh command fails, start it using the following appropriate method:
- On a local Sun or Solaris system, enter: /usr/etc/inetd
- On a local HP system, enter: /etc/inetd
For this operating system, check to see if the inetd daemon is active by locally entering lssrc -s inetd. If the inetd daemon is not active, start it by entering startsrc -s inetd. If the inetd daemon is active, refresh it using the following methods:
- On local installations of this operating system, enter:
```
/usr/bin/refresh -s inetd
```
- On remote installations of this operating system, enter:
```
rsh HostName /usr/bin/refresh -s inetd
```
- On local Sun or Solaris systems, enter:
```
ps -aux | grep inetd
kill -1 Process_id_of_inetd
```
- On remote Sun or Solaris systems, enter:
```
rsh Hostname ps -aux | grep inetd
rsh Hostname kill -1 Process_id_of_inetd
```
- On local HP systems, enter:
```
/etc/inetd -c
```
- On remote HP systems, enter:
```
rsh Hostname /etc/inetd -c
```
- Check that the DSMIT ports are active, enter:
```
netstat -Aan | grep 6125
netstat -Aan | grep 6126
```
  Output should look similar to this:
```
<HexNumber> udp 0     0     *.6125   *.*
<HexNumber> tcp 0     0     *.6126   *.*     LISTEN
```
- If you are using Network Information Service (NIS), add the entries into /etc/services and /etc/inetd.conf on the NIS server and rebuild the NIS maps. Refer to the NIS Overview in the AIX 5L Version 5.1 System Management Guide: Communications and Networks for further information.
- Check the /etc/inetd.conf file on both the server and clients for the following two lines:
```
rex dgram udp wait root /usr/dsmit/bin/crexd crexd -t30
rex stream tcp nowait root /usr/dsmit/bin/srexd srexd -i
```
  If these lines are not present in the file, add them to the file and refresh the inetd daemon.
- Check the /etc/services file on both the server and clients for the following two lines:
```
rex 6126/tcp #srexd port
rex 6125/udp #crexd port
```
  If these lines are not present in the file, add them to the file and refresh the inetd daemon.
- Ensure that the server name is accessible to the client system by issuing the following command:
```
ping Server HostName
```
  If the system returns the server's TCP/IP address, the network is working correctly. If it does not, add the server host name to the /etc/hosts.equiv or /etc/hosts file. If the system is using a name server, the problem could be the result of name server failure. See TCP/IP Name Resolution in AIX 5L Version 5.1 System Management Guide: Communications and Networks for more information on name servers.
- Sun, Solaris, and HP systems may report /usr/share/SMIT/configdb files not found. System configuration is required before you can use DSMIT management devices and TCP/IP or NFS subsystems.
- Sun, Solaris, and HP support require that the default shell contained by the /bin/sh file is the Bourne Shell.

If You Cannot Establish a Secure Session

If a message such as "establishing secure session failed" is displayed, refer to the following steps:

Be sure the system you are trying to establish a session with is defined as a managed machine. On the DSMIT server, do the following:
1. Type dsmit env.
2. Select List All Machines.
  OR
3. Type lsdsmitm.
  
  Note: If the machine you are trying to establish a session with does not appear in the list of managed machines, you should add it. See Modifying DSMIT Security Configuration for more information.
Install a new key on the managed machine:
1. Remove the managed machine.
2. Add the machine.
3. Install the new key.
  
  Note: See Modifying DSMIT Security Configuration for more information.

If You Cannot List All Members of A Domain

If the output from the lsdsmitd command does not include all of the hosts that were added to a domain, check the /usr/share/DSMIT/hosts file for spaces between client names. Remove all spaces in the definition of a domain.

If You Cannot Define A Working Collective

If you cannot define a working collective because you get the message:

All the hosts should have one and the same Operating System

Check that the clients included in the working collective have the same operating system.
If you are attempting to define a working collective that does not use this operating system, you must use the Redefine Working Collective Members menu to redefine the working collective. Do not include this local operating system in the working collective.

If You See Double Entries in the Domain Management Menus

Double entries can occur if you completed step 4 in the Database (Databases) section even though DSMIT was installed on the HostWithNewLevel. To correct this situation, remove the NewLevelDirectory and complete all the steps except step 4 in the "Database" section.

If You Cannot Connect to the File Server

If you get messages such as the following:

Cannot acquire exclusive lock

Cannot connect to the file server

Enter the dsmit-unlock and dsmit-dec commands on the file server.

If You Cannot Run dsmit-addkey on a Client

If you cannot run the dsmit-addkey command because you get the message:

dsmit-addkey: Problems with /tmp/new-v5srvtab -- exiting.

Be sure that you copy the correct new-v5srvtab file from the managing machine to the client. It should come from the /tmp/dsmit directory on the managing machine.
Be sure the hostname of the client matches exactly the hostname from the security configuration on the managing machine. For example M1.austin.ibm.com. must be written out in full; it can only be abbreviated as M1 if it is abbreviated as such on the managing machine.

If You Cannot Access/Open File /tmp.installp_cmd

If you get the Cannot access/open file /tmp.installp_cmd message when trying to install a software package, the version of installp on the remote machine is different than the SMIT database version on the local machine. To fix this, install the correct version of the database on the local machine. Refer to Databases .

If You Cannot Select the Next Screen

You may get the following message cause you cannot select the next:

There is a conflict in selecting the next screen - Make other selections in previous screens and/or 
exclude some hosts from the collective

The next menu to display is not exactly the same on all machines in the working collective. This indicates one of the following:

Dialog IDs are different for each machine
Number of options are different for each machine
Options are different for each machine.

For example, for the menu Change/Show Characteristics of a Tape Drive, the working collective is defined to be M1 and M2. On M1, a 4.0 GB 4mm tape drive exists and on M2, a 2.3 GB 8mm tape drive exists. When DSMIT attempts to display a menu, you will get the above error message. This is because of the different dialog type options. On the 4.0 GB tape drive, it has the extra option Use data COMPRESSION and on the 8.0 GB, it has the Use EXTENDED file marks option. DSMIT cannot display different dialogs at once. Each dialog must be the same to work.

To correct this problem, you must work with each group of machines that have the same dialogs by temporarily excluding the machines that are different and working with them later.

If You Cannot Get Solaris Machines Into Working Collective

If you cannot get Solaris 2.3 and Solaris 2.4 machines into the same working collective, check /usr/share/DSMIT/hosts to be sure all machines are listed as Solaris and not Solaris_2.3.

Related Information

Chapter 1, Distributed System Management Interface Tool.

Installing DSMIT.