Cannot post to nginx trac right now, I think the trac db index is corrupted, so will post here for now.

The http_stub_status module seems broken in 1.3.15 (was fine in 1.3.14)

We were using 1.3.14 without any problems, upgraded to 1.3.15 this past weekend and now see very strange surges in our statistics before they return to normal.

Investigated the output from http-stub-status-module and the numbers are erratic.
The accepts/handled/requests counters are broken.

Here are three reading done in a row, in less than one minute.
NOTE THE SURGE IN THE MIDDLE SET which obviously are incorrect numbers.

Active connections: 27
server accepts handled requests
782806 782806 1463305
Reading: 0 Writing: 1 Waiting: 26

Active connections: 32
server accepts handled requests
9391043 9391043 17594034 < < < < - - - - ? ? ? ?
Reading: 1 Writing: 1 Waiting: 30

Active connections: 67
server accepts handled requests
782915 782915 1463531
Reading: 0 Writing: 1 Waiting: 66