-
Type:
Task
-
Status: Done (View Workflow)
-
Priority:
Normal
-
Resolution: Done
-
Component/s: ics_fpsActor, ics_mcsActor
-
Labels:
We long ago, at ASIAA, confirmed that we could stuff MCS centroids into the database and retrieve them from FPS fast enough (and faster than going through a FITS file). I don't think we are doing as well at Subaru. In any case this should be checked.
@cloomis: has a test against a local disk been performed to confirm the original timings?
Better, but not good. Running with the actor in /tmp:
pfs@mhs-ics:~$ oneCmd.py mcs expose bias 2018-10-10T13:53:31.580 sent mcs expose bias 2018-10-10T13:53:33.621 mcs i exposurestate="exposing" 2018-10-10T13:53:33.631 mcs i text="ETYPE: bias" 2018-10-10T13:53:33.632 mcs i text="slice name: /data/mcs/2018-10-10/PFSC00007000_" 2018-10-10T13:53:37.241 mcs i exposurestate="reading" 2018-10-10T13:53:38.850 mcs i text="Time for exposure = 3.618860." 2018-10-10T13:53:38.851 mcs i text="Time for image loading= 1.607460."
There was a spurious copy of an uncompressed image to NFS; we have gained another 1.4s:
2018-10-16T00:40:09.714 sent mcs expose object expTime=1.0 2018-10-16T00:40:11.681 mcs i exposurestate="exposing" 2018-10-16T00:40:11.690 mcs i text="slice name: /data/mcs/2018-10-15/PFSC00009500_" 2018-10-16T00:40:15.533 mcs i exposurestate="reading" 2018-10-16T00:40:15.660 mcs i text="Time for exposure = 3.849244." 2018-10-16T00:40:15.660 mcs i text="Time for image loading= 0.125715." 2018-10-16T00:40:17.886 mcs i filename="/data/mcs/2018-10-15/PFSC00009500.fits" 2018-10-16T00:40:17.886 mcs : exposurestate="done"
Ok thanks!