monit speech alert proposal

[Martin Pala]

Hi,

recently i played with very interesting speech dispatcher 
(http://www.freebsoft.org/speechd.en). It provides device independent 
network interface for speech synthesis. It is provides very simple 
protocol called SSIP (speech synthesis independent protocol). The 
message is immediately reproduced by sound system. The simplest example:

unicorn:~/cvs# telnet 127.0.0.1 6560
Trying 127.0.0.1...
Connected to 127.0.0.1.
Escape character is '^]'.
SPEAK
230 OK RECEIVING DATA
Restarted Service apache

       Date: Thu, 04 Sep 2003 02:33:03
       Host: www.tildeslash.com

Your faithful employee,
monit
.
225 OK MESSAGE QUEUED

It is possible to define message priority (multiple clients can submit 
messages of different priority at once - dispatcher will manage all 
messages in correct order). You can chose the voice, rate, language - 
many languages are supported - it uses as backend output modules 
festival, flite, epos and mbrola (mbrola is not free), it supports sound 
icons and another very interesting features.

For example to choose language:
SET SELF LANGUAGE

To choose voice sample:
SET SELF VOICE MALE2

To set priority (|important|, |text|, |message|, |notification|, 
|progress|):
SET SELF PRIORITY MESSAGE

It has very simple client C API - it is possible to use it or write 
simple monit module without speechd library.

What about to add support for it, for example by extending alert statement:

# global section
set alert speechd:192.168.121.15:6560  # operators center - will 
receive all alerts with default format
...
alert speechd:192.168.121.100:6560 on { timeout } with mail-format { 
Alarm, service $SERVICE is down } # manager - short critical info

This way it could be possible to alarm one or more places (operators 
center, admins stations, [PR] managers, etc.) from all of your monit 
nodes - you can optionaly select some event types only (if you don't 
want to have all alarms in your speakers), override default message 
format and other things as usual in normal alert statement - the only 
difference is recipient specification.

You can find in the attachment record of result of above example 
(default format of monit alert message) in male and female synthetic 
voices (for fun :) i thing it sounds good ...

Martin

monit_female.mp3
Description: audio/mpeg

monit_male.mp3
Description: audio/mpeg