|
| 1 | +#### Prototype Devices for the transport between FLPs and EPNs |
| 2 | +-------------------------------------------------------------- |
| 3 | + |
| 4 | +#### General |
| 5 | + |
| 6 | +The devices implement following topology: |
| 7 | + |
| 8 | + |
| 9 | + |
| 10 | +- **flpSyncSampler** publishes timeframe IDs at configurable rate (only for the *test mode*). |
| 11 | +- **flpSenders** generate dummy data of configurable size and distribute it to the available epnReceivers. |
| 12 | +- **epnReceivers** collect all sub-timeframes (according to number of FLPs), merge them and send further. |
| 13 | +- flpSenders choose which epnReceiver to send a given sub-timeframe to based on its ID (`timeframeId % NumEPNs`), ensuring that sub-timeframes with the same ID arrive at the same epnReceiver (without need for additional synchronization). |
| 14 | +- Upon collecting sub-timeframes from all flpSenders, epnReceivers send confirmation to the sampler with the timeframe ID to measure roundtrip time. |
| 15 | +- epnReceivers can also measure intervals between receiving from the same FLP (used to see the effect of traffic shaping). |
| 16 | +- The devices can run in *test mode* (as described above) and *default mode* where flpSenders receive data instead of generating it (as used by the Alice HLT devices). |
| 17 | +- Optional deployment and execution via DDS. |
| 18 | + |
| 19 | +#### Device configuration |
| 20 | + |
| 21 | +The devices are configured via command line options. Most of the command line options have default values. These default values are for running in the *default mode*. Running in *test mode* requires a few modifications. |
| 22 | + |
| 23 | +These are the required device options: |
| 24 | + |
| 25 | +**flpSyncSampler** (only for the *test mode*) |
| 26 | + |
| 27 | + - `--id arg` Device ID |
| 28 | + - `--data-out-address arg` Data output address, e.g.: "tcp://localhost:5555" |
| 29 | + - `--ack-in-address arg` Acknowledgement Input address, e.g.: "tcp://localhost:5556" |
| 30 | + |
| 31 | +**flpSender** |
| 32 | + |
| 33 | + - `--id arg` Device ID |
| 34 | + - `--num-epns arg` Number of EPNs |
| 35 | + - `--data-in-address arg` Data input address, e.g.: "tcp://localhost:5555" |
| 36 | + - `--data-out-address arg` Data output address, e.g.: "tcp://localhost:5555" |
| 37 | + - `--hb-in-address arg` Heartbeat input address, e.g.: "tcp://localhost:5555" |
| 38 | + |
| 39 | +Default overrides for *test mode*: |
| 40 | + |
| 41 | + - `--test-mode arg (=0)` "1" to run in test mode |
| 42 | + - `--data-in-socket-type arg (=pull)` "sub" for test mode |
| 43 | + - `--data-in-method arg (=bind)` "connect" for test mode |
| 44 | + - `--data-in-address arg` In test mode, the signal from flpSyncSampler sends to this address (data is generated). |
| 45 | + |
| 46 | +**epnReceiver** |
| 47 | + |
| 48 | + - `--id arg` Device ID |
| 49 | + - `--num-flps arg` Number of FLPs |
| 50 | + - `--data-in-address arg` Data input address, e.g.: "tcp://localhost:5555" |
| 51 | + - `--data-out-address arg` Output address, e.g.: "tcp://localhost:5555" |
| 52 | + - `--hb-out-address arg` Heartbeat output address, e.g.: "tcp://localhost:5555" |
| 53 | + |
| 54 | +Default overrides for *test mode*: |
| 55 | + |
| 56 | + - `--data-out-socket-type arg (=push)` "pub" for test mode (pub will just discard the data because no receiver in test mode). |
| 57 | + - `--ack-out-address arg` For test mode provide address of the flpSyncSampler. |
| 58 | + |
| 59 | +To list *all* available device options, run the executable with `--help`. |
| 60 | + |
| 61 | +When running with DDS, configuration of addresses is also not required, because these are configured dynamically. |
| 62 | + |
| 63 | +Example for the *test mode* can be found in `run/startFLP2EPN-distributed.sh.in` for manual run, or `runO2Prototype/flp_epn_topology.xml` for DDS run. For *default mode* there is an example DDS topology in `../topologies/o2prototype_topology.xml`. |
| 64 | + |
0 commit comments