Experimental Physics and
Industrial Control System

Stephanie Allison (415)926-4534 <[email protected]> · Fri, 23 Feb 1996 8:52:08 -0800 (PST)

Matthias wrote:

>In addition it would help if any write action to the IOC's could be logged on a 
>file. This would help to keep track and analyse problems. This would need an 
>output channel from the CA-server on the IOC to a (i.e.) NFS-mounted disk.
>The output would look like:
>date time user@machine changed record from value to value

This must of been discussed in the past.  Sorry if this is a rehash.

I agree that we need a mechanism for logging changes by the user and
that the user, host, channel name, and old and new values need to be
in the message.  I think the logging should be by the client instead of the 
IOC CA servers.  If somebody is knobbing a setpoint (or using a slider or 
doing a correlation plot), we wouldn't want a message for each change.  
Instead, the knob client should log when the user first assigns a channel 
to a knob and perhaps some periodic time after that.  If somebody is 
restoring a backup (which may include hundreds of channels), the client 
should just log the name of the backup that got restored.  When the user 
does an individual text update, message, or choice button,  the change 
should be logged by the display manager.  But this should be configurable - 
this should be an option that the display creator can set.  It would be 
nice if at the time a controller is created, EDD provided a on/off field 
for logging changes.  If set, DM would send the change to some server or 
write to some file - configurable using environment variables.

++ Stephanie    [email protected]

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System