I have discovered a network peer that is definitely behaving badly. What makes this your problem is that it's apparently a Shareaza peer.
The bad behavior is quite simple: when queried, it returns a bunch of search results for files I want. I add them to the download list and three of them queue; the rest quickly go to "Searching (No Sources)", which is obviously wrong (not only is there a source, that source is online right now if I'm queued on it for three files!).
Then the queued ones advance to the head of the queue, fetch a TigerTree, and then ... Searching (No Sources).
If I re-add the source, three of them queue again, so it's not that the source suddenly went offline.
Similarly, if I re-do the search, the same source shows up with the same bunch of hits after a short time.
The source exists whenever I do a search. When files are requested, sometimes it exists and queues the requester and sometimes it doesn't. When anyone gets to the front of the line, suddenly it doesn't exist. Or so the pair of Shareaza instances (mine and the source's) claim, illogical though that is.
What is going on here? And how is it to be fixed so I can actually get these files? Someone's online right now and willing to share them. Ergo I should be able to get them.
His client reports itself to be Shareaza 2.1.something; mine is 2.5.2.0.
Is it a bug in 2.1 Shareazas that has since been fixed, or might this be a problem with even the current version?
There actually seem to be at least two bugs here. One is probably in the source's copy of Shareaza and apparently causes it to incorrectly slap a downloader with a busy signal when they hit the front of the line instead of actually serving the promised file. The other is a long-standing issue involving push sources that has been around for ages: a busy from a push source is treated as if it were a 404 rather than a busy, which is clearly incorrect. That bug is at the downloading end, i.e. it is present in my own copy of Shareaza 2.5.2.0.
It's possible the other bug is in 2.1.x and has been fixed since. I'm requesting that the drops-busy-push-sources bug that is definitely still extant be fixed too, and that someone verify that whatever's causing that source to respond with "busy" when a host reaches the front of its line was fixed in the past.