Jonathan, Dave:

As the first part of the upgrade of h1tw1 from its V1+RAID hardware (gentoo 2.6) to the all-in-one hardware (debian 8)  (already done for h1tw0) I did the following:

During a 5 minute break time on both h1tw1 and h1tw3:

stopped daqd running on h1tw1, took this out of monit control, we are no longer running daqd on this unit

switched h1tw3 to write to a new minute_raw directory, its old data is now ready to be deleted since its disk is at 96% full

We then reconfigured h1nds1 (the default NDS server) to temporarily:

Serve recent archived data (24sep - 21oct) from h1tw1 while these data are being transferred to LDAS-GW

Serve current data (15:37 todays onwards) from h1tw3 using a temporary NFS mount point.

Because the raw minute trend 5-minute data blocks were not synchronized between h1tw1 and h1tw3, there is actually an overlap of a minute between the two data sets. It appears that dataviewer is insensitive to the data duplication, but ndscope gives an error (unexpected number of channels were received).

For a work around tonight, if you wish to use ndscope to get minute trend data, please use h1nds0 as your NDS server, for example

NDSSERVER=h1nds0:8088 ndscope -t yesterday -t now H1:PEM-EX_TEMP_ROOF_WEATHER_DEGC

we will work on the fix tomorrow.