Experimental Physics and
Industrial Control System

"Jeff Hill" <[email protected]> · Mon, 4 Aug 2003 11:44:14 -0600

> When looking at the channel access client information on IOCs running
> EPICS R3.13.4 we usually see the same name listed multiple times.  
> Here is an example.
> 
> Client Name="bartlett", Client Host="d0ol20.fnal.gov", V4.8, Channel
> Count=6
>          TId=0X171E54, Protocol=TCP, Socket FD=11
>          Secs since last send   1.63, Secs since last receive   1.63
>          Unprocessed request bytes=0, Undelivered response bytes=0
>          Client's DNS Host Name d0ol20:55513 State=up
>          33358 bytes allocated
>          MUO_PROC_23/FD(0r-)     MUO_PROC_23/FD(0rw)
> MUO_PROC_23/MEM(0r-)
>          MUO_PROC_23/MEM(0rw)    MUO_PROC_23/CPU(0r-)
> MUO_PROC_23/CPU(0rw)
> 

This indicates that your client side application has created multiple channels
for the same PV name.

> Does this indicate a problem with our channel access client?  

This certainly indicates that there is room for optimization. Whether this is
necessary or not will depend on your situation - how many channels, how often
this gets used etc.

> What is the key for the code following the record names?
> 

This is documented in the ADG. 

The entries all take this form.

> MUO_PROC_23/CPU(0rw)

pvName(nnnn{r,-}{w,-})

Where the nnnn indicates the number of monitor subscriptions attached to the
channel. The {r,-} indicates whether there is or isn't read access. The {w,-}
indicates whether there is or isn't write access. Sorry about the cryptic code,
but compactness seemed to be appropriate when printing a list of channels (which
could be long). 

Jeff

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System