Present: Enrico, Roland, Giovanna, Milo, Eric, Frank
General discussion on impending data taking
documentation: idea is to have two FELIX specific pages (a fair amount of work still needs to be done):
status:
- new libfabric working, no network issues remaining
- using numactl
- f/w: issues with links starting still exist
plan to start now in standard partition 1
- changing partition causes the WIBs to reset (as their timing changes), and a flx-init is then necessary to recover
- idea is that shifters should not (in general) have to deal with this
data validation:
- we should strive to look at other information than merely noise numbers: error information, timestamp & convert count consecutiveness, etc.
- (Milo:) this should be largely possible with the now-existing online monitoring scripts; can be automated?
re-ordering & compression: not yet fully integrated (Roland will attempt to devote some "spare" time to this)
- need to decide how to determine whether fragment data have been compressed or not; preference seems to be to indicate this in the fragment header (not using fragment type)
- plan is also to reduce now-useless header information
- leave timestamps & convert counts in for now
Compression note
- comments welcome
- scope limited to honouring our obligation to μBooNE to provide a document showing how we've used their data
- should then continue with a note that is more complete and relevant to ourselves
Next meeting: will decide early next week whether or not to meet on Wednesday, August 29
There are minutes attached to this event.
Show them.