What needs to be done to replace the DAQ Doctor
Created by: hsakulin
1) Additional checks
-
High individual FED Deadtime -
High Partition Deadtime -
High global deadtime (with analysis of breakdown (individual issue #46)) (*) -
(individual issue #17 (closed)) Check that physics stream rates are within limits (configurable, different sets for different machine modes, may vary over time, exceptions for special runs, maybe tie to HLT Key or L1/HLT Key). ( finer check in the future: limits may depend on HLT menu, inst. luminosity, prescale-column ) -
(individual issue #28) Temperature checks -
(individual issue #29) (check all other checks that used to be done by the DAQ doctor)
(*)
-
Use DeadTimeBeamActive for runs with beam in the machine (INJECTION PROBE to STABLE BEAMS) -
Use DeadTime for cosmic runs (BEAM DUMP to SETUP)
2) New data sources
2a) New data source: Filter Farm Monitoring
-
(individual issue #47) meet with Srecko to find out if it is always there -
(individual issue #17 (closed)) HLT output rates per stream, stream sizes -
(individual issue #44 (closed)) CPU usage per machine -
Fill level of RAM disks (individual issue #32), output disks (individual issue #45)
2b) New data source: errors from XMAS (spotlight)
2c) Where do we get the temperature monitoring from? (individual issue #28)
II=> 3) Audio alerts
-
LHC machine/beam mode change -
DAQ state change -
Subsystem going to RunningDegraded / RunningSoftErrordetected / Error -
(individual issue #43) L1 or HLT rate out of limits -
High dead time -
No rate when expected etc.
4) Add support for audio
-
NM sending alerts to CMS-WOW in control room (test on office PC) -
(individual issue #35)Receive audio message from DCS, WBM, DQM (see: https://svnweb.cern.ch/trac/cms-daqdemo/browser/DAQTools/Monitoring/trunk/DaqDottoressa/RequestHandler.pm?desc=1) External message also create a notification. -
(individual issue #40) Tool for expert at home to listen to audio messages (in browser ? ) -
(individual issue #39) Add global dead time to plots. Use DeadTimeBeamActive all the time. -
(individual issue #41) 6) Searchable archive