Skip to content

Latest commit

 

History

History
77 lines (51 loc) · 2.73 KB

README.trigger.md

File metadata and controls

77 lines (51 loc) · 2.73 KB

Dadatrigger

This code is written for the AA-Alert project. (c) 2017 Jisk Attema, Netherlands eScience Center, ASTRON

Based on the PSRdada software.

Selectively dumping parts of a data stream with dada_dbevent

Setup:

  1. Main ringbuffer (that is being filled by fi. the fill_ringbuffer program). This also serves as a cache; ie it should hold on to data until it is out of the dumping window
  2. dada_dbevent. Listens to triggers on a network port (localhost:30000 by default), and when triggered copies data to the disk buffer
  3. Disk ringbuffer. Stageing place to write to disk
  4. dada_dbdisk. Dumps all data in the disk buffer. Prepends the header block.

Triggering

A trigger is a text file containing:

  • N_EVENTS followed by an integer number indicating the number of events in the trigger
  • On the next line, the start time of the observation in UTC, in the form YYYY-MM-DD-HH:MM:SS

then per event a single line with (space separated):

  • Event start as YYYY-MM-DD-HH:MM:SS
  • seconds fraction for start time as floating point number
  • Event end as YYYY-MM-DD-HH:MM:SS
  • seconds fraction for end time as floating point number
  • DM floating point number, not used further.
  • SNR floating point number, not used further.

or:

  • QUIT

A trigger can be send using fi: cat trigger | ncat localhost 30000

Output

Events overlapping in time are merged to a single event. The header block from the main ringbuffer is then prepended to the datastream, and the buffers are dumped to disk. Events are stored one per file.

The to map times to bytes, the BYTES_PER_SECOND is used together with UTC_START. These must be defined in the main ringbuffer header. The resulting byte index then rounded to the neareset multiple of RESOLUTION (also from the ringbuffer header). It is rounded down for start, and up for end times. We use this to make sure we only dump complete ringbuffer pages.

Usage

  1. Setup the ringbuffers with the setup_buffers.sh script
  2. Get data flowing by running setup.sh
  3. Trigger an event by sending message to the dada_dbevent program (see the explanation above, and the generated file trigger)

Replaying an event

The event (ie. the dumped data) can be replayed, for further processing.

  1. create a ringbuffer with exactly the same page size
  2. connect a dada_diskdb to the ringbuffer as writer and pass it the file(s): dada_diskdb -k -f [-f ...]
  3. connect your processing script to the ringbuffer as a reader

Notes

  1. dada_dbdisk can only accept 1024 input files. Configure dada_dbdisk to write bigger files if you run into this limit. Set FILE_SIZE to 384000000.0