I wonder...
Since yesterday I started seeing a large amount of new blocks propagating on Testnet, a somewhat disturbing amount of new blocks all with one single transaction. And they keep coming with only seconds in between. I haven't studied the content of the blocks, but that's easy for anyone who wants.
A similar thing happened not long ago. If you take a look at
http://blockr.io/, click on Charts and click in the left on the Testnet coin (just under bitcoin, number 2 from above) and check the graphs, you will see a similar spike around April 14th (edit: direct link here:
http://tbtc.blockr.io/charts). The current has started to decline in frequency at the time of writing this.
Since yesterday my self-developed node, which currently does not propagate anything, merely consumes everything it is handed from the network (it is Testnet after all, yeah? And I'm running it in a debugger). I received approximately 20,000 (yes, twentythousand) new blocks on testnet in a timeframe of around 24 hours. All neatly hashed with appropriate difficulty, as far as I can see. So yes, I wonder.... what is going on here?
I think your problem might well arise from the fact that since you also, probably, received thousands of new blocks within a very short timeframe, but actively propagate (forward) these to other Testnet nodes, you might trigger a ban-filter for too frequent updates. I don't know, I have never looked at the "official code" so cannot say for sure, but that's how I would implement it too, i.e. banning nodes that update me suspiciously frequently.
You might not be the only one having this problem.
This of course leads me to another question (or three): How come this is at all possible? Who has this sort of hashing power? What's happening?
I mean, I'm connected to a rather large amount of peers and changing them frequently for new addresses i receive. And each time I take the (debug) node down and up again, it starts over with the very few seeds I have. So I should have a pretty good variety, that is, not being fed blocks by only one malicious node.
Any ideas? Do I have a bug in my software that I'm unaware of?
