IBM: TCP/IP for VM

Getting a VMDUMP of select TCP/IP virtual machines

The following methods can be used to obtain various types of dumps of select TCP/IP server virtual machines.

Notes:

For the several of the commands that follow, CP TERMINAL default LINEND and ESCAPE characters — the cross-hatch (#) and double quote ("), respectively — are assumed to be in effect.
The dump recipient user ID for all commands illustrated here is the user ID: TCPMAINT

To obtain a dump immediately for a given server, using the server user ID itself:

Logon the given virtual machine (for example, TCPIP)
Issue:
#cp vmdump 0-end dcss format cms to tcpmaint
After the dump is created, disconnect from the server console:
#cp disconnect

To obtain a dump using commands issued from the TCPMAINT user ID:

For the TCP/IP server virtual machine (TCPIP):
- To obtain a dump immediately, issue:
  netstat cp vmdump 0-end dcss format cms to tcpmaint
- To set up a trap to automatically get a dump when an exception occurs, issue:
  netstat cp trace prog xx cmd vmdump 0-end dcss format cms to tcpmaint"#trace end all netstat cp query trace
  where xx is one of the following hexadecimal interrupt codes:
  04 - Protection Exception 05 - Addressing Exception
  This will take a dump at the point of the failure and will then turn tracing off. If you would like the virtual machine to be IPLed after tracing is turned off, append "#IPL CMS PARM AUTOCR to the end of the NETSTAT command string shown above.
  Note that the " is the ESCAPE character and the # is the LINEND character. Issue the CP Q TERM command to see how they are defined on the system where this command will be issued and substitute those values.
For a pool of SSL server virtual machines:
- To obtain a dump immediately, issue:
  ssladmin system cp vmdump 0-end dcss format cms to tcpmaint (sslserver all
- To set up a trap to automatically get a dump when an exception occurs, issue:
  ssladmin system cp trace prog xx cmd vmdump 0-end dcss format cms to tcpmaint"#trace end all (sslserver a ssladmin system cp query trace ssldadmin closecon
  where xx is one of the following hexadecimal interrupt codes previously described

The CP QUERY TRACE commands cited above allow one to confirm the trace setting is correct.

For the TCP/IP server, the trace setting is reported in the response to the NETSTAT command used to process the query command within the server.

For SSL servers, the server console data must be reviewed, because the system command responses are not returned to issuer of the SSLADMIN SYSTEM command. This review can be done after the consoles are closed via the SSLDADMIN CLOSECON command (cited above), and the relevant console file is available in the TCPMAINT virtual reader.

Processing The VMDUMP After It Has Been Taken

After the dump has been taken, it needs to be processed before it is supplied to IBM support for analysis.

Log on the virtual machine where the dump was sent (for all of the commands For these example commands, this is: TCPMAINT

Process the dump as follows:

Define a temporary disk that will be large enough to store the dump:
```
  define t3390 5ff cyl 100
  format 5ff t
  1
  tmp5ff
  
```

Order the virtual reader and load the dump:


 query rdr * all               (To identify the spool ID of the dump file)
 order rdr yyyy                (yyyy is the dump spool ID; this command places the dump
                                file at the beginning or the virtual reader)
 dumpload spool outfile pmxxxxx t   (where xxxxx is a 5-digit PMR number)
 yes

The above commands will generate a dump file that you then need to COPYFILE (PACK (to preserve the file content and structure), and then FTP in binary form to an IBM drop site for analysis.

For detailed instructions on supplying the dump file to IBM, consult the information an this URL: