So it seems that this very high read rate was a singular event. I wish I had looked a bit earlier at this bug: in order to figure out which requests might have caused this high read rate I tried to look today at the access logs for 2010-10-12 -- but we keep them only for 3 week or so :(
Running "grep hwsubmission *" on the DB reports (devpad: /home/lpqateam/ public_ html/dbr/ daily) returns:
db-report- 2010-10- 04.txt: hwsubmissiondevice || 4505.06 tuples/sec 2010-10- 08.txt: hwsubmissiondevice || 0.20 tuples/sec 2010-10- 12.txt: hwsubmission || 527430.31 tuples/sec
db-report-
db-report-
So it seems that this very high read rate was a singular event. I wish I had looked a bit earlier at this bug: in order to figure out which requests might have caused this high read rate I tried to look today at the access logs for 2010-10-12 -- but we keep them only for 3 week or so :(