- Table of contents
- 6/8/2015 Notes
Rob, Cristiana, Serdar, Leon, Craig G., Alec, Jan, John, Eric, Andrew and Martin (taking notes)
- Andrew points out that the limit of DAQ stability is now at 94 buffer nodes.
- Andrew suggests to report the data Ganglia.
- Ganglia knows what buffer nodes are in the partition.
- Python makes this particularly easy.
- Therefore we should just run Rob's program on every buffer node and not have any aggregation.
- Perhaps we could broadcast some JSON data.
- Scripts could start at boot or the DAQ could do it.
- DAQ TDUs do this, so we can follow their examples.
- TCR does not report to Ganglia, but it could be added.
- Spill Server reports to Ganglia.
- I wonder whether we can extract the information upstream in the trackers to get the slice to track association somehow.
- Andrew suggests that we compare the pointers themselves since they are the same.
- Rob points out that the TDC values in the Track object are floats. Andrew says that they should be integers. Rob will find out how that float TDC value is constructed.
- Andrew suggests that we use one of the unused farm nodes (farm-11) to run Eric's tests.
- Alec suggests that we use the high energy trigger information to skip over things that are too "messy".