Opened 6 years ago
Last modified 5 years ago
#2683 assigned defect
Tool Shed stats corrupted?
Reported by: | Tristan Croll | Owned by: | Greg Couch |
---|---|---|---|
Priority: | moderate | Milestone: | |
Component: | Tool Shed | Version: | |
Keywords: | Cc: | ||
Blocked By: | Blocking: | ||
Notify when closed: | Platform: | all | |
Project: | ChimeraX |
Description
Looks like something weird has happened during the time the Tool Shed was down. Both Clipper and ISOLDE show massive spikes in downloads during the period (a few thousand apiece). Other bundles show similar spikes. In addition, the "star" ratings have changed dramatically for seemingly all bundles. Before the down period ISOLDE had < 10 votes with an average 5 stars, now it's 74 with an average 2 stars. Clipper has gone from (as far as I recall) zero votes to 49 with an average of 1 star.
Change History (6)
comment:1 by , 6 years ago
follow-up: 2 comment:2 by , 6 years ago
Right now I’m most concerned about all the 1-star ratings. I’m in the middle of writing up an application for a group leader position in Australia, and that could have a real, serious impact on my prospects.
comment:3 by , 6 years ago
Were there many ratings prior to the outage? If not, you can just delete all the bogus ratings. If not, it will have to fall in the same bucket as the downloads.
follow-up: 4 comment:4 by , 6 years ago
Oh - didn't know I had the power to delete them. Just went ahead and deleted all ratings for ISOLDE and Clipper - there were only 7 or so on ISOLDE and none on Clipper last time I checked anyway. Might want to do the same for all the other bundles - they all seem to have been affected in the same way. On 2019-12-13 22:07, ChimeraX wrote:
comment:5 by , 6 years ago
Need to write django management command to delete all ratings from a particular IP address and/or to see if a particular IP address has made lots of rating in a short time frame, and delete them.
comment:6 by , 5 years ago
Owner: | changed from | to
---|
My guess is that during certificate renewal, UCSF IT scanned the sites, resulting in massive "download" counts. I'm not sure how to deal with this, but will look into whether individual download requests are saved in the database. If so, I can just delete the ones that came from the scanning host and we should get "real" stats back. If not, I can only suggest that we reset the counters.