Tags:
create new tag
view all tags

CDS Notes

CDS greps

To check the health and efficiency of the higher level code, we should check the icuevents files. They can be found on msslje at /home/itos/archdir.

This means

grep -i timeout PKT*/*icuevents.log
grep -i ERR PKT*/*icuevents.log
grep -i blocked PKT*/*icuevents.log
grep -i IDLE PKT*/*icuevents.log

and looking at all the errors.

We're interested specifically in:

  1. going to blocked_failsafe (IDLE) because of problems like bright fields, software timeouts etc.
  2. observing efficiency (time spent not observing)
  3. errors which don't produce pages (which in the past have been concerning!)

Also

grep -i safepoint PKT*/*icuevents.log

The reason being it is interesting to know how much time we wasted in safepoint when we could have been in safepoint doing flatfields.

grep -i "Count Rate" PKT*/*icuevents.log

"Count Rate" means the star cat stopped us looking at something. The number of GRBs that get lost because of this is a very interesting efficiency measure especially since the starcat has too many duplicates.

Also if PTs are stopped, the planning was wrong.

grep -i slewabort PKT*/*icuevents.log

Slewabort normally means that the ICU didn't respond to a slew request quickly enough which violates the prime directive.

fgrep "drift" PKT_*/*icuevents.log

A new part of the CDS work is now to look for small differential and absolute drifts in the icuevents logs as they won't be so easy to spot now.

fgrep "exhausted" PKT_*/*icuevents.log

Look for "mode exhausted" as they should not happen.

The automated lists in http://msslje.mssl.ucl.ac.uk/lists/ may be useful.

nhklis.txt     - nhk messages ranked
nhklisbig.txt  - nhk messages with codes ranked (bigger)
rtslis.txt     - RTSs ranked
errlis.txt     - ICU errors
timeouts.txt   - timeouts that may need explaining
countrate.txt  - high countrates for statistics
slewabort.txt  - slewaborts (probably caused by ICU FSW)
blocked.txt    - which may need explaining
icuevents.txt  - sorted/no duplicates ICU events

icuevents.txt can also safely be used for filter wheel stats and safe/idle time stats as it is in order and doesn't have duplicates.

CDS Results

There's some new CDS results

Spacecraft slew settling accuracy:

http://www.mssl.ucl.ac.uk/~pjs1/swift/cds/att/

Pretty good recently. Good for optimizing the choice of window size, star catalogue search radius etc.

Real count rates in each filter:

http://www.mssl.ucl.ac.uk/~pjs1/swift/cds/counts/

Note that this is from 10s engineering housekeeping data and shows a point for Earth angle or filter against count rate every 10s.

Filter wheel moves:

http://www.mssl.ucl.ac.uk/~pjs1/swift/cds/fw/

Note that the number of filter wheel moves increased quite a lot for January. Maybe all the GRBs.

Edit | Attach | Watch | Print version | History: r10 < r9 < r8 < r7 < r6 | Backlinks | Raw View | More topic actions
Topic revision: r10 - 2006-02-21 - HowardHuckle
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback