Should we impose an arbitrary upper limit on the number of requests running at any one time? It might help to smooth over temporary resource shortages. I believe mrogers' token passing algorithm includes an element of this, but we don't need to wait until then to implement it.
If we have too many requests running, we wouldn't accept any remote requests, or start any local ones. However, this may be overengineering? We have a variety of overlapping feedback mechanisms already, some of which don't seem to be interacting properly - for example, pre-emptive bandwidth allocation should keep down bwlimitDelayTime, but it doesn't - or does it? -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20060919/3188b1be/attachment.pgp>
