This morning some models went from RCG3.2 to RCG3.2.3. Changes made in RCG3.2.1 mean that when an updated model is ran with mx_stream-3.2 (the code which sends the model's data to the DAQ), the keep-alive message coming back from the DAQ Data Concentrator is not interpreted correctly and the front-end DAQ bit is erroneously set bad.

The data from the model to the DAQ is perfectly correct, and the status of the data flow as determined by the Data Concentrator is correct.

There are three options:

1. make no changes, accept that the CDS overview will have red daq bits for some models

2. for models running 2.3.2, do not show the DAQ bit on the overview

3. for any front end with at least one model running 3.2.3, upgrade all the models on the computer to 3.2.3 (including IOP) and run mx_stream-3.2.3

I don't want to do 1), this can lead to red-fatigue when viewing the overview.

Option 3 is intrusive and will require customization of the monit script which starts mx_stream in order to start the correct version depending upon the computer name. It will require upgrading all models on h1susb123, h1seib2, h1susex and h1susey (and possibly on h1asc0 or downgrading h1asc model to RCG3.2).

For now I'm implemented option 2). A new composite fec-status MEDM called H1CDS_FE_3_2_3_MICRO_CUST.adl has been created and used for the models h1sus[i,e]tm[x,y], h1isibs, h1asc.

I tested these options on h1pemmx. It is currently running 2.3.2 on IOP and h1pemmx, and mx_stream-3.2.3.

I have reset the build release pointers back to RCG3.2 for now.