ezca write process reports a change in EPICS record in guardlog, but doesn't actually write the value
Exposed this issue using lines 256 - 263 in ^/trunk/isi/h1/guardian/SEI_ENV.py (happens both before and after the sleep(0.1) we installed superstitiously) (see context in LHO aLOG 58672):
We found intermittent problems with setting EPICs variables via the guardian's ezca interface. When writing to a bunch of similar channels with a for loop, for one or two of the channels, the guardian log would say one or two got written, but the value never got written (and the rest did).
We exercised the CONSTRUCTION state 3 times, and each of the three times we got a different result: (1) ITMY ST1 and ITMX ST2 actuator allowable counts were not set to max, (2) only ITMY ST1 was not set, and (3) it all worked as written. Each time in (1) and (2), the guardian log reports a change in variable, but MEDM and ndscope reveal that the value was never changed.
Anecdotally, Jenne and Sheila report past experience with this too.
@thomas-shaffer can provide a bit more explicit detail.