Re: Problems using devSnmp from NSCL

["John A. Priller" <priller@frib.msu.edu>]

The first thing I notice is that the "reply mask" from your INP field is 
set to INTEGER:, but that won't work parsing the reply string of 
"INTEGER: off(0)" because "off(0)" isn't an integer.  Try using '(' as 
your mask:
field( INP,  "@$(CRATE) %{WCR}outputSwitch.$(CH) ( 100")

Whether that's what's causing your problem with dbpf to the .PROC I 
don't know, but it is a problem.
On 07/02/2014 01:47 AM, Florian Feldbauer wrote:
> Hey,
>
> yes both records show up, when I run the `dbl` command.
> Also there is no error message at initialization.
>
> epics> snmpr 2 PANDA:LMD:MUPIX:LV:TEST:ReadSwitchOnOff
> ============================================================
> devSnmp report
> ============================================================
> read task loops : 3034376 (49.8 per sec)
> send task loops : 3034301 (49.8 per sec)
> active requests : 0
>
>   ========================================
>   host wienerpl506
>   ========================================
>   SNMP version    : SNMP_VERSION_2c
>   Max OIDS/req    : 32
>   Get Queue items : 0
>   Set Queue items : 0
>   Active sessions : 0
>
>     ----------------------------------------
>     Group 'guru'
>     ----------------------------------------
>     Host             : wienerpl506
>     Send count       : 56804
>     Reply count      : 56804
>     Error count      : 0
>     Avg reply time   : 0.0 msec
>     Best reply time  : 0 msec
>     Worst reply time : 20 msec
>     PV count         : 34
>     OID count        : 20
>
>     PVs:
>
>     ----------------------------
>     PV 'PANDA:LMD:MUPIX:LV:TEST:ReadSwitchOnOff'
>     ----------------------------
>     OID name         : WIENER-CRATE-MIB::outputSwitch.u0
>     Host/Community   : wienerpl506/guru
>     Set type         : 's'
>     Special flags    : 0x0000
>     Settings count   : 0
>     Max data length  : 100
>     Raw value string : 'INTEGER: off(0)'
>     Reply mask       : 'INTEGER:'
>     Value string     : 'off(0)'
>
>
>     OIDs:
>
>
> verifying devSnmp tasks are running...
>   sendTask is alive
>       sendTask_abort  : 0
>       sendTask_exited : 0
>       globalLastTick  : 2014/07/02 08:37:21
>   readTask is alive
>       readTask_abort        : 0
>       readTask_exited       : 0
>       readTask_inSelect     : 0
>       readTask_fds          : 18
>       readTask_block        : 0
>       readTask_ignoreBlocks : 0
>       readTask_timeout      : 9 sec 995935 usec
>
> I've just noticed, that the record seems to update correctly but with 
> a READ_ALARM.
> The record bo/longout combination for setting 
> WIENER-CRATE-MIB::outputSwitch.u0 is working fine...
> Concerning the mb/bi/o/direct records: for thinks like the switch 
> on/off stuff it would be nice...(since you don't need the additional 
> longin/out record anymore)
> originally I also wanted to use mbbiDirect for the outputStatus, but 
> this has 20 bits, so in this case one needs the longin/calc records 
> anyway...
> Best regards,
> Florian
>
>
> On 07/02/2014 03:20 AM, John A. Priller wrote:
> > I might be a little slow helping debug it, I'm away from work until 
> > July 10.  Do the PV names you're trying to access show up in a 'dbl' 
> > command?  If so, what shows up when you have the snmp driver do a 
> > dump of them?  For example:
> > snmpr 2 PANDA:LMD:MUPIX:LV:TEST:ReadSwitchOnOff
> >
> > will do a level-2 debug dump of PV/oid info.
> >
> > The reason we don't support mbbo/i/direct (yet, anyway) is simply 
> > because the old snmp driver we based on version on didn't support it 
> > and we've never yet needed it.  It's something we could add if 
> > there's demand for it (and time is available, you know how that goes).
> > On 07/01/2014 09:00 AM, Florian Feldbauer wrote:
> > > Hey John, hey all,
> > >
> > > we have a Wiener PL506 low voltage power supply in our lab 
> > > controlled via SNMP.
> > > I tried using the devSnmp module from NSCL 
> > > (https://groups.nscl.msu.edu/controls/)
> > > Nearly everything is working just fine, except for two records:
> > >
> > > record (longin, 
> > > "PANDA:$(subsys):$(dev):LV:$(sector):ReadSwitchOnOff") {
> > >   field( DTYP, "Snmp")
> > >   field( INP,  "@$(CRATE) %{WCR}outputSwitch.$(CH) INTEGER: 100")
> > > }
> > > record (longin, 
> > > "PANDA:$(subsys):$(dev):LV:$(sector):ReadRegulationMode__" ) {
> > >   field( DTYP, "Snmp" )
> > >   field( INP,  "@$(CRATE) %{WCR}outputRegulationMode.$(CH) INTEGER: 
> > > 100" )
> > > }
> > >
> > > $(CRATE) is the IP-Address of the PL506, %{WCR} = "guru 
> > > WIENER-CRATE-MIB::"
> > > and $(CH) is "u0".
> > >
> > > When I try to process these records, I get the following error:
> > >
> > > epics> dbpf PANDA:LMD:MUPIX:LV:TEST:ReadSwitchOnOff.PROC 1
> > > filename="../dbTest.c" line number=375
> > > No such file or directory - dbPutField error
> > >
> > > Any idea what's goning wrong here?
> > >
> > > Another question to John: Is there a particular reason why e.g. 
> > > mbbi/o(Direct) or bi/o records are not supported by devSnmp?
> > > Best regards,
> > > Florian