Reports until 12:49, Tuesday 28 November 2017
H1 CDS
david.barker@LIGO.ORG - posted 12:49, Tuesday 28 November 2017 - last comment - 14:33, Tuesday 28 November 2017(39545)
long baseline minute trend dataviewer plots drawing issues

Chandra noticed some strange dataviewer plots when asking for long look-back minute trends (see attached). After some investigation, I found:

The problem is with h1nds1's configuration of archived raw minute trends. The most recent archive spans the dates 5/26 to 10/18. The older archives are not being accessed, which I think is a result of my reordering the archived directories in the daqdrc file. After Dan and I had moved all the minute trends to their own RAID, I noticed that raw minute trends are ordered oldest-to-most-recent which would not be efficient since most data requests are for recent (within the last year). I reordered the list to most-recent-to-oldest and now only the most-recent directory is being read. When we are able to restart the NDS, I'll reverse the order to verify this. In the mean time, minute trends before 5/26/2017 are unavailable.

Images attached to this report
Comments related to this report
david.barker@LIGO.ORG - 12:54, Tuesday 28 November 2017 (39546)

opened FRS9506 for this ticket.

keith.thorne@LIGO.ORG - 14:33, Tuesday 28 November 2017 (39550)CDS, DAQ
The daqd process assumes that the list is from oldest to newest.   This is how it has been since Alex Ivanov implemented it.  Yes, likely need to improve documentation but usage has been limited to LLO and LHO.