If you are trending slow channels which are acquired by the EDC spanning the time of the EDC+DAQ restart yesterday (11:30 Tue 28 Jan 2025) and you are using the default NDS (h1daqnds1) you may see outlier data for a period of about 10 minutes. This is because we staggered the DAQ restart to preserve front-end data, starting with the 0-leg and EDC first, and then the 1-leg later after fw0 was back to writing frames. 

The EDC change yesterday was to add/remove mechanical room vacuum channels, and the vacuum systems are at the beginning of the EDC DAQ configuration file. The result of this is that after h1edc was restarted at 11:28, from h1daqfw1's perspective almost all the EDC channels moved relative to each other (channel hop), meaning that for each channel nearby channel data was being written instead for about 10 minutes.

h1daqfw0 does not have this issue because h1edc was restarted while it was down, such that when it started writing frames again it was running with the new h1edc configuration. Ditto for raw minute trends on h1daqtw0.

Switching your NDS from the default h1daqnds1 to h1daqnds0 removes this issue. The attached plots show a 24 hour trend of the outside temperature in degC obtained from nds1 and nds0. The nds1 command is:

ndscope --light -t "24 hours ago" -t now H1:PEM-CS_TEMP_ROOF_WEATHER_DEGC

and the nds0 command is

ndscope --nds h1daqnds0:8088 --light -t "24 hours ago" -t now H1:PEM-CS_TEMP_ROOF_WEATHER_DEGC