Experimental Physics and
Industrial Control System

"Allison, Stephanie" <[email protected]> · Wed, 26 Jun 2002 17:57:23 -0700

To anybody who uses EZCA (EZ Channel Access) Monitors:

During testing of a new CA client that uses EZCA for monitors,
we came across a unique (rare) problem:

(1) While the client was stuck in ezcaDelay (1 sec timeout), 
    a new value came in for a PV right before the IOC was rebooted.  
(2) When the client then called ezcaNewMonitorValue, it was told
    a new value was available (the needs-reading flag was set for
    that PV).
(3) The client then called ezcaGetWithStatus and was NOT
    given the new value and was told (via error return) that the 
    channel was disconnected.
(4) After the next ezcaDelay, the ezcaNewMonitorValue still
    returned TRUE so the client again called ezcaGetWithStatus 
    and got the same error (and did this over and over again,
    logging each failure at the 1 second rate until the reboot 
    finished).

Perhaps using EZCA monitors is not a good idea.  In any case,
I would like to change ezca.c in one of three ways, (1) being
my preference:
(1) For monitors only, if the needs-reading flag is set,
    the various EZCA get routines return the available value 
    without checking for channel disconnect and then 
    reset the needs-reading flag.  Next time the error is 
    returned.  
OR
(2) The various EZCA get routines reset the needs-reading
    flag if the channel is disconnected and the last value
    is lost.
OR
(3) ezcaNewMonitorValue returns FALSE if the channel is
    disconnected and the last value is lost.

Comments/other suggestions?  Thanks very much,

Stephanie Allison ([email protected])

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System