skip book previous and next navigation links
go up to top of book: HP OpenVMS Alpha Partitioning and Galaxy Guide HP OpenVMS Alpha Partitioning and Galaxy Guide
go to beginning of chapter: The Graphical Configuration Manager The Graphical Configuration Manager
go to previous page: GCM Server Log Files GCM Server Log Files
go to next page: PerformancePerformance
end of book navigation links

Troubleshooting the GCM Server  



Any time that the GCM server is started using the normal startup procedure, it is running in a mode that supports automatic restart behavior. When run in its normal mode, the GCM server produces its run-time log file, but offers little additional information for troubleshooting possible GCM server problems.

The following sections contain information to assist you in troubleshooting possible GCM server problems.

Obtaining Diagnostic Information  

At times it may be beneficial to run the GCM server in a mode that allows some diagnosis to occur. Diagnostics produce procedural trace information that is either written to the GCM server event log, if enabled, or output to the screen, if enabled. Typically, diagnostic output is directed to the screen; however, the event log output is useful if you need to forward it to HP support personnel for help.

To get diagnostic screen output from the GCM server, you must run the GCM server interactively rather than as a detached process. To do this, stop the GCM server if it is currently running, and restart it from a DECterm window by entering the following command:

$ RUN SYS$SYSTEM:GCM_SERVER.EXE
Once the GCM server is running, connect to it from a GCM client with administrative privileges, and edit the GCM administration database as follows:
  1. Select Edit Admin Database from the File menu.
  2. Select the Server Init page.
  3. Locate the Diagnostic text field and note the buttons for selecting the desired output.
  4. Click the Screen output button and enter the appropriate diagnostic code or codes, according to the following table. You can set more than one flag at a time.

    Code (decimal) Code (hex) Function Action
    0
    0
    DIAGNOSTIC_DISABLE
    Disables diagnostics.
    1
    1
    HEARTBEAT_TRACE
    Servers disconnecting.
    2
    2
    TRANSACTION_TRACE
    General troubleshooting.
    4
    4
    XML_TRACE
    General troubleshooting.
    8
    8
    LOCK_TRACE
    Server hangs.
    16
    10
    COMMAND_TRACE
    Disables execution and dumps packet.
    32
    20
    CRYPTO_DISABLE
    GCM administration database troubleshooting.


NoteBy default, regardless of whether the GCM client-server communications is using security (SSL), the GCM administration database is always encrypted. On rare occasions, you may want to disable encryption of the GCM administration database GCM administration database file. By setting the CRYPTO_DISABLE flag, GCM servers will no longer encrypt and update the GCM_ADMIN.EDB file. Instead, they will output plain ASCII XML to GCM_ADMIN.DAT, and will accept input on startup from GCM_ADMIN.DAT. This allows you to directly edit the XML structures within the GCM administration database and to review modifications made by the GCM client and server.


NoteRun the GCM servers with the CRYPTO_DISABLE flag set only for troubleshooting because running the servers with this flag set exposes GCM user authorization records in simple ASCII unencrypted form.

Remember to disable diagnostics before returning a GCM server to service. If you do not disable logging, the related GCM server log files can grow very large and GCM server performance will suffer.


Potential Problem Areas  

Heartbeats are a potential problem area for heavily loaded systems. For GCM servers to detect the presence and loss of GCM clients and GCM servers in other instances, periodic heartbeat transactions are issued. The GCM server is smart enough to optimize the heartbeat process, altering the rate based upon usage, and requiring only one-way heartbeats. At times, GCM server or network activity delays the issuance of a heartbeat. The GCM server is designed to tolerate a number of missed heartbeats before assuming disconnection.

You can tune heartbeat values to suit specific needs by using the GCM client to edit the GCM administration database. Three values apply: the Client Pulse, Server Pulse, and Flat Line. The Pulse values are the number of seconds between heartbeat transactions. The Flat Line value is how many heartbeats can be missed before disconnection. These are preset to values that yield good behavior without undue traffic. Under special circumstances, you can alter these values if you find your sessions are being disconnected. These values have no effect on the overall responsiveness of the GCM with regard to command and event processing.

Timeout Detection  

The GCM client contains a debug hook to turn off timeout detection entirely. To enter a debug mode, press Ctrl+Alt+D+B. A TEST menu item indicates this mode is active. In this mode the GCM client remains connected regardless of how long it takes to get a response back from a slow GCM server (such as one that is being debugged).


go to previous page: GCM Server Log Files GCM Server Log Files
go to next page: PerformancePerformance