Jim, Dave:

Sheila and Jenne made h1asc filter changes over the weekend and we noticed today that the diff file for this system was filled with many copies on a single filter. Investigating further we noted that the problem was with the three filters ASC_CHARD_P, ASC_CHARD_P_A and ASC_CHARD_P_B. Jim noted that these consist of a core name, and filters with additional suffixes, so we suspected a potential string matching issue. In the case of h1asc, the running H1ASC.txt file in the chans/tmp directory had 638 copies of the ASC_CHARD_P filter module, with varying header names.

We were able to reproduce the problem on the DAQ Test Stand. We made changes to the three filters and loaded them individually. When the "core" named filter (ASC_CHARD_P) was loaded many copies of this were created. Further individual loads compound the problem, which is why h1asc ended up with so many copies.

A bug has been opened for this: https://bugzilla.ligo-wa.caltech.edu/bugzilla3/show_bug.cgi?id=1014

We scanned all the H1 filter modules looking for filters which are core names to additional filters. We found 65 cases in the 8900 total number of filters. These filters are distributed over the following systems:

h1alsex
h1alsey
h1asc
h1calex
h1caley
h1iscex
h1iscey
h1lsc
h1oaf
h1odcmaster
h1omc
h1psldbb
h1pslfss
h1psliss
h1pslpmc
h1susetmx
h1susetmy
h1susitmx
h1susitmy
 

Until this issue is resolved, we recommend that users do not individually load filter modules and instead perform full filter file loads from the GDS_TP MEDM screen.