scheduled announces trigger unscheduled scrape requests

[gunzip]

i've noticed that when the announce and scrape intervals are different, for example, the announce interval is 35 min and the scrape interval is 30 min, the following happens:

1) at the 30-min mark a successful scrape occurs, and then schedules the next scrape in 30 min (OK)

2) at the 35-min mark a successful announce occurs (OK), but this triggers yet another scrape even though it was just scraped 5 mins earlier and wasn't scheduled for another 25 mins.

for those cases where people are running a high number of torrents this extra scraping doesn't provide anything useful and just adds unnecessary overhead. it seems a better solution is to let the scrape and announce intervals update independently.

also, the scheduled scrapes are reported to the logfile, while unscheduled scrapes are not. but the Web-client > Inspector > Tracker does show these are in fact occurring.

Linux Debian, transmission-daemon 2.30b2 (12354)

[jordan]

What's happening is that the announce response was a superset of a scrape response, so we count it as both an announce and scrape and reset the timers for both. This should result in less bandwidth used on scrape, not more. :)

[gunzip]

OK jordan, thanks for the explanation.