Page History
...
- cnf starts dab/eb ports (everyone gets CM well-known ipaddr/port via cnf. port corresponds to platform number)
- (needs work) drp/eb nodes find their own IB addr's
- user asks for partition, sends out PLAT msg (says "who exists?")
- all drp/eb nodes reply with their name/hostname
- CM sends ALLOC message to nodes selected for partition, e.g. 4 drp, 2 db. (says "who is participating in partition")
- eb/drp open up appropriate ephemeral ports (possibly depending on who else is in partition) and report them to the CM (response to ALLOC)
- CM sends CONNECT Messages to eb/drp nodes with assigned pseudo-random ID numbers and connection information. ID numbers could also be communicated with ALLOC message.
- eb/drp nodes connect
- (needs work) need to add CONNECT_COMPLETE msg to send to CM
- eb/drp/control.py do their usual mainloop
...
DTI Collection Information
Collection behavior of DTI (should be implemented in the as-yet-to-be-named "blob" process).
On PLAT:
- epics prefixes
- available upstream/downstream ports
On CONNECT:
- alloc ports and connect upstream/downstream
Some details. The epics prefixes are:
- "where" (DAQ:LAB2)
- XPM:crate#
- DTI:crate:slot
DTI port info (UDET devices, e.g. Jungfrau):
- crate/slot/port/device (device auto-detected with PGP)
XPM port info (TDET devices, e.g. hsd, tool, tes)
- crate/port/device
Issues:
- need upstream/downstream ids for detectors/KCUs
- downstream ids are 32-bit and should communicate IP address (lower bytes), firmware type, card#
- multiple detector auto detect in 1 TDET-style KCU doesn't work trivially
- KCU firmware is detector specific
- TTOOL/TES don't currently have device detection
Overview
Content Tools