Scalable RIP diagnostic tool (central logging)

This page applies to Harlequin v13.1r0 and later; and to Harlequin Core but not Harlequin MultiRIP

Scalable RIP processes (controlling RIP + all farm RIPs) send log messages in a publish / subscribe fashion to a central logger, which must reside on the central blade (i.e., logging is done from all the Scalable RIP processes to a central logger). If a central logger is not running and therefore not listening the messages get thrown out). Scalable RIP central logging is typically used for watching detailed progress, an aid for the diagnosis of problems and acts as a useful debug tool to see what is happening amongst all the cooperating processes in a single output stream.

A central logger is provided with HHR and is run as so:

C:\HQNRIP\tools\ripfarmdiagtool.exe -o C:\temp\ripfarmdiagtool.txt 2019-02-18 10:27:28.354: 192.168.132.133 fl-000 INFO: Process

starting...

2019-02-28 10:27:28.354: 192.168.132.133 fl-000 INFO: Invoked as:

C:\HQNRIP\tools\ripfarmdiagtool.exe -o C:\temp\ripfarmdiagtool.txt 2019-02-28 10:27:28.354: 192.168.132.133 fl-000 INFO: Listening on

tcp://*:9876

Command-line options are:

Option

Description

-o <filename>

--output

<filename>

Output log messages to the specified filename.

Example: -o C:\temp\ripfarmdiagtool-output-01.txt

-a

--append

Instead of overwriting the specified output file, append to it instead.

This option takes no arguments and must be used alongside -output or -o

Example:

--append --output C:\temp\ripfarmdiagtool-output.txt>

or --a --o C:\temp\ripfarmdiagtool-output.txt

or -a -o C:\temp\ripfarmdiagtool-output.txt

If this option is omitted, the output file will be overwritten.

Verbosity of RIP logging

In the ClientLogging section of the global config file, the following filters get applied to every thread in the Scalable RIP. Sending a lot of messages has an overhead so we do client-side filtering, which can be controlled with the following section in the global config file:

TEXT

                      "#": "RIP farm and Scalable RIP client logging options.", "ClientLogging": {
                        "#": [
                      "Turn on or off which types of log message will be sent to the central logger.",
                      "Sending verbose messages can cause a lot of network traffic but is very",
                      "useful for diagnosing problems. SendError, SendWarning and SendInfo should be used under",
                          "normal operation."
                        ],
                        "SendError": true, "SendWarning": true, "SendInfo": true, "SendDebug": true, "SendVerbose": true

Keys are explained in the following table:

Name	Default	Description
`SendError`	true	Send log messages marked as error to the central logger. MON_ERROR_CRITICAL and MON_ERROR
`SendWarning`	true	Send log messages marked as warning to the central logger. MON_WARNING_FAILURE and MON_WARNING
`SendInfo`	true	Send log messages marked as info to the central logger. MON_INFO_CONTROL, MON_INFO, MON_INFO_JOB
`SendDebug`	false	Send log messages marked as debug to the central logger. MON_DEBUG
`SendVerbose`	false	Send verbose logging to the central logger. MON_DEBUG_VERBOSE + every single ZMQ message that sent between every single thread in the Scalable RIP or RIP farm. This can generate a fair amount of logging, but it is incredibly useful to see what is happening between all the threads and processes.