Experimental Physics and
Industrial Control System

Emmanuel Mayssat <[email protected]> · Wed, 18 Mar 2009 11:33:32 -0700

On Wed, 2009-03-18 at 11:42 +1100, Andrew C. Starritt wrote:
> Hi all,
>  
> On a couple of IOCs, in the st.cmd file I start a number (7 or 9) SNL 
> programs (well actually each IOC it is the same SNL program – with
> option +r 
> and parameterised).
>  
> Sometimes one or more random-ish instance fails to connect to 
> all the PVs and essentially hangs. If I seqStop and restart the
> hung sequence program it runs as expected okay.

In your initialization state, you should check for PV connections:

ss turnOnOffStateSet {
  state turnOnOffInitializationState {
    entry{
      seqLog("--> turnOnOffInitializationState\n");
      p_deviceName=macValueGet("deviceName");
      seqLog("deviceName=%s\n", p_deviceName);
    }
    when (pvConnectCount() == pvChannelCount()) {
      seqLog("All channels connected\n");
      delay(5);
      seqLog("Reseting communication\n");
    } state resetState
    exit {
      seqLog("<-- initializationState\n");
      /* Buttons are intialized */
      resetRqstdBO=0;pvPut(resetRqstdBO);
      turnOffRqstdBO=0;pvPut(turnOffRqstdBO);
      stndbyRqstdBO=0;pvPut(stndbyRqstdBO);
      highVltgOnRqstdBO=0;pvPut(highVltgOnRqstdBO);
    }
  }
 [....]
}

>  
> I have some questions which I hope can get help with.
>  
> The output from seqShow is:
>     Program Name         Thread ID  Thread Name          SS Name
>     sncControl_Selection 0x32a75a0  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32a2d70  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32a4580  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32a5d90  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32aa5c0  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32abfe0  sncControl_Selection cycle_magnets
>     sncControl_Selection 0x32a8a30  sncControl_Selection cycle_magnets
>  
> Is there an easy way to tell which instance is which?

That's an interesting question...
I am not sure you can do this through seqShow.
You can do this with the log file (seqLog) but it may be a little
tedious.

> Can I some how parameterise the program name, thread name or SS Name?
> I tried:
>  
>    program sncControl_Selection{NAME}
>  
> and
>  
>    ss cycle_magnets{NAME}
>  
> in the .stt file but neither would even compile.

In stt file:
program snovaTurnOnOff ("deviceName=(null)")

In ioc startup file:
seq snovaTurnOnOff,"deviceName=${SNOVA_DEVN},logfile=
${LOG_DIR}/${SNOVA_DEVN}-snovaTurnOnOff.log,debug=1"

>  
>  
> I also observed that the initial hang-up can is avoided by adding some
> sort of delay between the seq calls, e.g.:
>  
> seq (sncControl_Selection,"NAME=SR03ID01, ILK_NAME=SR03II01")
> dbl * > /tmp/xx
> seq (sncControl_Selection,"NAME=SR04ID01, ILK_NAME=SR04II01")
> dbl * > /tmp/xx
> seq (sncControl_Selection,"NAME=SR05ID01, ILK_NAME=SR05II01")
> dbl * > /tmp/xx
>  
>  
> I have seen this on both epics base 3.14.8.2 with seq-2.0.11 on 4 year
> old
> redhawk Linux system and using epics base 3.14.9 with seq-2.0.11 on a
> newish
> CentOs 5.1 system. Any thoughts as to what is happening? 
>  
> If the only solution is a small delay, is there a standard IOC shell
> sleep 
> command I could use instead?

In iocsh Try: 
help
help epicsThreadSleep
epicsThreadSleep(5)

>  
>  
>  
> Regards
> Andy
>  
>  
> Andrew Starritt, Principal Controls Engineer, Australian Synchrotron
> 800 Blackburn Road, Clayton, Victoria 3168
> ph: (03) 8540 4164 , fax: (03) 8540 4200
> mailto:[email protected],
> http://www.synchrotron.org.au
> 
>  
>  
>  
> This message and any attachments may contain proprietary or
> confidential information. If you are not the intended recipient or you
> received the message in error, you must not use, copy or distribute
> the message. Please notify the sender immediately and destroy the
> original message. Thank you.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System