Last modified: 2014-05-13 20:42:16 UTC
I spoke with <Flexman> this evening since he was having some trouble with his mobile app. The app requests thumbnails of various heritage-protected buildings for displaying to a user. The requests to thumb.php are proxied via a single server as an anonymous user. Flexman is hitting the ping limiter but doesn't feel he's requesting 70 thumbs in 30 seconds (or 700 in 30 for standard thumbs). I can't find any evidence in fluorine:/a/mw-log/limiter.log (or the archives) that his IP is hitting the limits. We're going to do some live testing tomorrow so I can tail things in real time. Putting this bug in for tracking so I don't forget.
Fixed by operations today with Gerrit change #133124. Esams upload proxies were using IPv6 and were hitting the rate limiter. Went back to IPv4 and the issue is resolved. More generally, getting the IPSet work into deployment so we can use CIDR ranges well help this sort of thing from cropping up in the future.
