Opened 14 years ago
Closed 14 years ago
#1739 closed Enhancement (fixed)
1.50b[1,2] dump poorly-performing peers even when we don't have candidates to replace them
Reported by: | charles | Owned by: | charles |
---|---|---|---|
Priority: | Normal | Milestone: | None Set |
Component: | libtransmission | Version: | 1.42+ |
Severity: | Normal | Keywords: | |
Cc: |
Description
Summary says it all. 1.4x had a bug that kept nonproductive peers for too long. That's fixed in 1.50b[1,2] but those betas go too far in the other direction -- they cut peers even when replacement candidates aren't available.
Instead of being too slow or too fast to cut peers, the "Just Right" solution should keep 1.50b[1,2]'s bugfix, but check for replacement candidates *before* cutting the ones we've got.
Peers in an error state should be unconditionally closed.
If we're a seed or partial seed, then peers that are also seeds or partial seeds should be unconditionally closed.
Change History (2)
comment:1 Changed 14 years ago by charles
- Status changed from new to assigned
- Type changed from Bug to Enhancement
comment:2 Changed 14 years ago by charles
- Resolution set to fixed
- Status changed from assigned to closed
Note: See
TracTickets for help on using
tickets.
fixed in trunk in r7761, r7763, and r7765.
backported to 1.5x branch in r7767.