TRACE has been created at the request of a number of our customers and is new with Cohort V3. TRap and Act on Console Events traps console messages on a system and checks those messages against a series of filters to determine if action should be taken. To help centralize system management TRACE also includes a mechanism for trapping console messages and forwarding them to another node, if desired. The node may be in the same cluster OR remote. The message forwarding will work as long as the nodes are networked. If TRACE is running on the node the message gets forwarded to it can be configured to detect the event and take action there.
The product has been installed and in use at a Canada wide organization for a number of months. They run small to medium size clusters in a number of cities in their WAN. With TRACE they trap the console messages at the smaller sites and forward them to a larger site where their OpenVMS system managers can examine the messages and respond appropriately. They use TRACE in conjunction with Hitman, but it will work with any 3rd party or inhouse application that sends console messages. A typical application for TRACE in their corporation:
1. At a smaller center Hitman detects that the system disk on Node Site5 is getting full. Hitman submits a batch job to purge the disk and will continue to send consoles messages until the condition is fixed. A filter can also be setup to catch the low system space message generated by VMS when a system disk gets to full.
2. TRACE detects the message. In order to give the purge job time to run and clear up the problem TRACE has been configured to ignore the first 3 messages. When the console message is detected the 4th time the message is forwarded to the remote node Manag1, a different node in one of the larger centers. The filter gets reset every hour; if the message is still being sent by Hitman the process will be repeated.
3. At the larger center a console message on Manag1 indicates that the system disk on Node Site5 is getting low on disk space. TRACE finds a match with a filter and then submits the specified batch job.
4. The system manager signs on and connects to Site5. She does some additional cleanup on its system disk and everything is now fine.
By trapping and filtering console events TRACE allows you to build batch jobs for a number of system events and submit these automatically. TRACE can be configured to log all matches against a filter, only action procedures it submits or both. TRACE can also be configured to filter all console messages or just those ones that were sent to certain operator classes.
Here’s the TRACE filter from node Site5:
Type : Action
Match : *hitman*sys$sysdevice*
Case Sens. : NO
Description : Hitman message re: system disk at Site5
Node : Site5
Command : TRACE_COM:SEND_TASK_MESSAGE.COM 1
Queue : site5$batch
User : system
Keep Log : no
Print Log : no
ange : 4:4
Reset After : 0 01:00:00.00
History : Log Submits: Yes
Log Matches: No
1 The send_task_message.com procedure is provided with Trace and will forward the message to a different node.