|
Hitman
HITMAN is the OpenVMS idle process and event monitor. HITMAN detects various conditions on your OpenVMS system and automatically takes action. You have total control on what actions are to be taken.
Idle Process Management
HITMAN is a very flexible idle process management system. With HITMAN you can:
 Specify system-wide idle and warning times
 Specify different idle thresholds by Username, Process name, UIC, Terminal, Server and Rights identifier
 Specify which processes to protect by Username, Process name, UIC, Terminal, Server and Rights identifier
 Specify the action to take when terminating a process (Force Exit, Delete Process or Both)
 Customize messages for warning and process termination.
 Detect "almost" idle processes with BIO, DIO and CPU thresholds
 Indicate which mode you want HITMAN to operate in (watch, warn, or hit)
 Indicate which types of processes HITMAN monitors for idleness (interactive, network, etc)
 Specify various modes of operations with DECWindows.
 Specify user defined actions to be taken when a user is idle
 Log off any or all users when BACKUP must run whether or not they are idle
 Specify prime and nonprime hours with different configurations for each.
Event Monitoring
HITMAN is also a very powerful event monitoring utility. When an event is detected you can send the console a message, mail a user a message and/or submit a batch job to perform any action required. With HITMAN you can:
 Monitor for Runaway processes
 Detect Processes when they stop or start
 Detect when someone runs a specific image
 Monitor for elevated priorities
 Monitor for processes with specific privileges
 Monitor disk space
 Detect device errors
 Detect stopped or stalled queues
 Ensure required jobs are in batch queues
 Detect missing processes
 Detect when a node leaves your cluster
 Detect processes in unusual states
Some new features added to Hitman for V8 or V9 are listed below. New features in Version 9.1 and V10 and listed on the V10 release notes page.
 The maximum number of processes that HITMAN can monitor has been increased to 2,000. We also have kits built and on-line for larger sites.
 HITMAN has a new audit file (HITMAN_DAT:HITMAN_AUDIT_FILE.node) that contains time and user information. HITMAN will write out to this file every time someone makes a change to the HITMAN parameters. It will also indicate if that change is the creation of a new permanent data file or modification of existing prime or nonprime data.
 The maximum number of events has been increased to 512.
 The inactivity thresholds (CPU, BIO and DIO) can now be specified for individual images. Many sites have reported situations where images/applications had background usage that caused them to appear to be active. Now HITMAN can be configured to kill these users without changing your default inactivity thresholds.
 A new process state event has been added. Now HITMAN can report by mail or OPCOM message when n or more processes are in a particular state. HITMAN can, optionally, submit a DCL batch job as well.
 HITMAN/VERSION now gives a multiple line display including the version and some license information.
 A new image count event has been added. Anytime HITMAN detects n users running a particular program it can be configured to send mail, send OPCOM messages and/or submit an action procedure.
 The action procedure for an event can be configured to run more than once. Many users have requested the ability to configure HITMAN events to submit the action procedure either every time HITMAN sees the event true, or for a user defined number of instances.
 A new utility program has been added that can send a forced exit to a different process. This can be used in conjunction with the user exit routines to issue forced exits to processes from batch jobs before they are stopped.
 All events have a new user parameter which is available in the action procedure as P8. This parameter can be used to pass site/event specific information to the action procedure.
 HITMAN's default working set parameters have been increased to help reduce paging. The new qualifiers /WSDEFAULT, /WSEXTENT and /WSQUOTA can be used to refine them further. The values can also be changed from within the menu.
 The process id (PID) is now passed to the action procedure for present process, priority and privilege events.
 A new resequence events function has been added. Use either /RESEQUENCE or the menu option under modify events, to resequence events from 2 on and fill in any gaps in the current event numbers. Event number 1 is the default event.
 The event mass change facility can now perform mass changes to any event field.
 A new monitor node event has been added. When set up it will send mail, send OPCOM messages and/or submit an action procedure every time node x notices that node y is no longer in the cluster. With V9 this event was also extended to include remote nodes that are not part of the cluster.
 A new repeating event has been added. This event can be configured to submit an action procedure every x minutes. This is useful for resetting license counts or submitting jobs that should be run periodically throughout the day.
 A new HITMAN/DUMP=EVENT command has been added. It creates a report to show all the events and their current status (whether or not HITMAN has seen this event as having occurred and been acted on).
 The HITMAN detached process now sets itself so that it cannot be swapped out. On heavily used systems HITMAN was occasionally getting swapped out and this caused significant delays in data collection.
 Users can now limit the number of processes that HITMAN will terminate in a single data collection pass. This can be used at sites where a large number of users become idle at roughly the same time to help spread out the process termination overhead.
 With V8 any valid device specifier that can be used with a GETDVI call can have a device error event configured for it.
 A new HITMAN/DUMP=SUMMARY has been added.
 The new /command qualifier can be used to create a DCL command procedure that will recreate your parameter file. This makes distributing a parameter file to multiple nodes that aren’t clustered easy and possible by e-mail. It also makes it possible to use a normal text editor to make significant changes to your parameter file by
updating the command procedure and then recreating your parameter file.
 The new /summary qualifier can be used to generate a summary of any HITMAN log file. The summary breaks down the contents of the log file by event type and, where appropriate, username.
 V7.x of OpenVMS includes a new DecWindows user interface. Some code changes have been made to support pausing or killing these sessions.
 It is now possible to use Hitman to monitor remote nodes to ensure they are available. Both the absent process and present process events can now be configured with a remote node name. Adding an absent process event on a system critical process, such as SWAPPER, for a remote node will cause Hitman to take action anytime that node goes down. There is some setup involved for these events on the remote node; the monitor remote nodes chapter of the user manual documents how to take advantage of this new feature.
|