Health Check Monitor:
process

Here is a technical description of the process steps at QC Garching required to refresh the Health Check (HC) Monitor daily.

Each operational machine which processes data of an VLT/VLTI instrument supported by the process has a cronjob creating and updating the Health Check plots.

This cronjob has four main components:

• a cronjob workflow tool [autoDaily]
• a scoring tool [scoreQC]
• an opslog parser [qc1Parser]
• a plotter [trendPlotter]

This workflow tool is called once per hour. It downloads all headers of the current date, creates reports, and downloads any new calibration data. These are exposed to an association engine which creates ABs (association blocks) which are the fundamental processing unit for QC (they correspond to the RBs or sof's). These ABs are then pipeline processed. Next, QC1 parameters are extracted and ingested into the QC database. Finally, the tool scoreQC derives scores based on configured thresholds. They help to assess the quality of the products.

Any failed AB is attempted to be processed one hour later. Failures can occur because of data are not yet available in NGAS, or for other reasons.

Scoring is a central concept for QC: products are quality-checked by evaluating their QC1 parameters and derive scores. They bare based on configured thresholds.

These thresholds can be fixed values (e.g. specifications), or relative values (like +/- 10% around the mean which is calculated from the trending), or statistical values (like +/- 3 sigma around the mean). Scores are very efficient in reducing information to its relevant core. Scores are aggregated into a final score for each product.

In addition, the Paranal ops log files are to be included in the HC plots. This is provided by the parsing tool [qc1Parser] reading the latest N Paranal ops logs where N is 2-3 days. The parsed QC1 parameters are stored in a local text file. From here they are available for plotting by the QC tool trendPlotter. That data source is called OPSLOG within the trendPlotter configuration.

The next and final step is to plot the QC parameters. The graphical information is by far the most efficient way for monitoring the instruments trends.

There is usually one trending plot per QC item. Find an example here.

The HC reports come in two varieties, the graphical view and the score-based view (quick-look).

The PSO data points are generally coded blue, to distinguish them from the data points produced by QC Garching. Only the N latest PSO data points show up.

The plots are converted into png format and fed into the web site http://www.eso.org/HEALTH.

Information about the content of the plots is documented here.