Automatic error handling?

[tonysab]

Hi, we are running on 6.5, does anybody know if there is any automatic error handling built in to this version that will highlight and report on mercator errors in maps, including Failed maps? At the moment we run a regular process which sweeps log files looking for errors.
Thanks in advance.

 

[OllyC]

I'm not sure in which versions these appeared, but there is the snmpagent which can be called from a management tool to display map failures, connection errors, etc.

Also there is the management console (a Mercator tool) which can display the same information in GUI.

Olly

 

[tonysab]

Thanks for the info, I am not sure what these are though, snmpagent and the management console? Can you elaborate, thanks.

 

[janhes]

Management console comes with 6.5 base product. SNMP agent is an additional product.
Best solution is analysing the log files - see AUDIT examples supplied with product.
It depends on what you want to do when a map fails.

 

[tonysab]

Thanks, yes we have a script which runs every 15 minutes and issues finds and greps to search for error text within the log files, and to email the details to our support guys. The problem is we are generating so many log files that the script is taking 30 minutes to run! We are therefore thinking it would be better to notify automatically whenever an error is generated, if that is possible.
I would be interested to see the audit examples, any idea where these would be? Thanks again.

 

[janhes]

Best to use a map which triggers on the creation of a log file. See AUDIT example for how to handle the log file.
We use this method to process around 10,000 logs per hour. Any errors are emailed to a support group.

 

[OllyC]

Simple Network Management Protocol (SNMP) is a way of broadcasting information on the state of your server. Applications like Tivoli, HP Open View will support it. So, if you want to get information on the Mercator Event Server status this is what you should use.

If you're interested in individual messages then use the audit logs (we use both checking of audit logs and the SNMP agent).

Rather than write a script to do finds and greps, use a Mercator map which pulls the out return code of the map. If its non-zero then send an email to the support team.

cheers,
Olly.

 

[janhes]

Audit example in Windows is C:\Mercator6.5\Examples\AUDIT\tryaudit.mms

 

[maltu]

In some places, the audit log is written to a simple 10 column audit table logging key info(mercator comes with a auidt log type tree, the map should be a 1-2 hour development job) - this is event driven.

You could hit the table every 5 mins or 15 mins to check for logs with map failure info and generate notifications and flag the columns that have been notified on.

should be close to what you are looking for?

 

[thisismondweep]

Hi,

One way we use to notify TX errors is by writing only errors to a file. There is a main DataStage TX map which calls all utility and transformation maps. The main map checks to see if any errors were produced in the subsequent steps; if yes, then it writes to a customised error file.

The advantage of this is that the error file is customised and so logs only error information. A event driven utility could then poll this file and easily notify, whenever a new entry is written to this file. I hope this helps.

If yes, let me know and I will provide you with more information about how this can be achieved.