Activity
We need to double check the timestamp field to ensure that it is really the measurement time, but otherwise this should be as simple as comparing the measurement timestamp to the timestamp from the current consensus.
We may want to SETCONF both FetchDirInfoExtraEarly and FetchDirInfoEarly to ensure we get a consensus every hour, so that we know if the measurement was after it was published, and we know the most recent value.
Adding this to the list of experiments for the tech report, though we may not get to it till right before the camera deadline.
Trac:
Actualpoints: N/A to N/A
Points: N/A to N/A
Parent: N/A to #2769 (closed)#2550 (moved) is potentially going to change how this works, because of the delay in choosing an advertised value..
This is now running on the chicago vm (which votes to urrass). See https://gitweb.torproject.org/mikeperry/torflow.git/blob/pid_control:/NetworkScanners/BwAuthority/README.spec.txt section 3 for details.
I included a fix for #1984 (moved), as well.
We'll see how it goes.
Trac:
Cc: N/A to karstenSpent quite a bit of time on this. It appears to be running fine on one bw auth. I created #4425 (moved) to monitor it and deploy on the others.
Trac:
Summary: Bw Authorities should apply ALPHA multiplier only if measurements change to Implement PID control feedback for bw auths
Status: accepted to closed
Actualpoints: N/A to 26
Resolution: N/A to fixed