[widar-wg] [Fwd: Re: Alerts - CM - configuration failure]

[Bill Sahr]

-------- Original Message --------
Subject: Re: [widar-wg] Alerts - CM - configuration failure
Date: Fri, 03 Jun 2011 12:50:15 -0600
From: Bill Sahr <bsahr at aoc.nrao.edu>
Reply-To: bsahr at aoc.nrao.edu
Organization: NRAO
To: Sonja Vrcic <sonja.vrcic at nrc-cnrc.gc.ca>
References: <4DDFEDF4.1000705 at nrc-cnrc.gc.ca>

Sorry to be late to the party.  Just a few comments, given below.

Bill

Sonja Vrcic wrote:

> 
> VCI configuration failure (for CM, VCI messages are incoming messages):
> 
>     * Action: Kill the script.
>     * Basically, this alert will be raised every time when CM generates
>       vciNack (Negative Acknowledgment) or vciReject (those messages are
>       described in the VCI Protocol Specification).
>     * Alert generated for every failed configuration. On the operator
>       screen, a new alert will overwrite previous alert (I presume that
>       all will be archived). 

If I understand Rich's schema correctly, an alert will be archived only
if the optional element "archive" and it's associated required attributes
are included in the alert message.  For the case of a software element
such as the CM agreed upon conventions will have to be established for the
values for archive attributes such as host, dev, & mon.  Rich, in one of
the examples given in his XML Alerts schema proposal gives the following:

<archive host="widar" dev="cm" mon="configuration_failure" />

> 
> Board configuration-failure (for CM these are outgoing messages):
> 
>     * Action: Write report.
>     * CM generates alert when 'send configuration to CMI' fails. CM log
>       will contain the cause of the failure: communication failure, CMIB
>       rejected the configuration (error), CM failure, etc.

Is the implication here that if an operator or other individual, upon
seeing an alert, wants further information re the cause of the alert
one will have to go to the CM log to obtain that information?

>     * In the case of communication failure, CM will periodically
>       re-transmit the message. The alert will not be re-transmitted.
>     * CM will clear alert when configuration is accepted by the CMIB
>       (board).
>     * Action: Write report.