Lech wrote a short introduction about using GrandLib as a readme in github. There is also a note by Jolan in the forge. An effort should be made to upgrade, merge, unify these docs.
Inserting simulations in the official repository and database will be done after simulations reprocessing.
Changing the run number when extra changes was discussed in the operation comitee, but it's not clear if it was implemented. François will try to check it and Lech and Olivier should discuss it during the next trip on site.
Inventory of suspicious fileds in the data was done and discussion about how to correct it started. This point will be discussed during the next trip on site.
Some useless data will be removed from the database by François.
Archiving the raw data is implemented. All raw data are automatically pushed on the tape (with double copy) system @ccin2p3 on a month basis (on 15th of a month, all raw files from the previous month are tared in a structured archive and pushed on the tape backup system).
After reprocessing the raw files older than 6 month will also be gziped on /sps to save space.
Some fields in Tadc (notch_filters_no_ch and samples_in_baseline_ch) are not well described (no type) in the Root format reference document (https://box.in2p3.fr/s/ipmk4XZP87pjRnr). François and Lech will need to check it.
François and Lech proposed a new du_id numbering at the hardware meeting in dec 2024. The main idea was to have a unique du_id over the whole experiment. This should be achieved by coding du_id on 7 digits, the first one indicating the site (1 for gaa, 2 for gpxxx,...) and the 6 others would be a serial identifying the DUs. We also propose to add more "information" in this numbering. A proposal would be to attribute consecutive ids by cluster of communication (i.e. DUs targeting the same rocket should have consecutive number). Another idea would be to also group serials by antennas position in the layout (infill vs coarse). François will try to prepare a proposal. If accepted, this numbering algorithm should be implemented in the antennas management software on the main DAQ.
Up to now, all the events were recorded into the database. But due to the high number of events (200 000 000 in the last 1.5 year) this is propably not sustainable on the long term. We thus think about removing systematic registration of individual events from the database and move to only register "interresting" events (i.e. reconstructed so of analysis level L2 or L3). François will investigate on that.
We plan to register models (RF chain, Antennas models,...) into the database. François will investigate on how to do it and how to provide an "easy" way to register models.
In any case, these data should be accessible only by the collaboration members
We will try to reprocess all the data with the last gtot version (with corrected GPS's rad to deg conversion precision and new format based on directories instead of single root file), if possible before next summer. However, Lech needs to correct a last bug in gtot (duplicates du) before.
As Grandlib is aimed to work only with this new format, we plan to replace all data in /sps/grand/data @ccin2p3 by the new reprocessed ones. However, we will keep old format root files in irods if someone needs it.
François will investigate if it's possible to keep the current database for "old files" and set up a new one for the reprocessing or if we simply replace the current database and website by the new one.