I hope you don’t mean that nodes will accept any garbage as SpentBook
record
Validly signed valid transactions So you need the correct keys
However if you try to fill the spentbook it would be akin to filling bitcoins address space or any asymmetric crypt keypairs. There are just too many, it’s 2^256 sets of keys.
i.e. the attack is similar to
- Generate a bitcoin key
- Check balance
3 Profit
Which means work both for client for generation and for node (Adult?) for storing it.
Benchmarks are needed to see, which will take more resources.
I know what 256 bits collisions mean
There is a good video about it:
Thx for the update Maidsafe devs
Always great to see the bug fixes @Maidevs
@mav 4 wielding an lightsaber
Keep hacking super ants
How different do you see that from storing any data, in terms of Adults?
No tokens are spent.
So such attack can last longer than attack, which requires constant inflow of money.
Also it may require more validation steps for Adult.
Will spentbook entries trigger section split by the way?
I understand that pure PUTs or GETs may be more dangerous.
But evaluating spentbook-related part of the system may be useful too.
Not checking for tokens is less work, not more though.
They are small, so not any more than storing that one more single wafer thin byte would
Seing as we are talking about talking to people - maybe we should talk to these folks as soon as bandwidth permits.
Resuming normal operation after attack stops is important characteristic of high quality software.
I suspect that voting for offline status may make network pausing hard to implement.
It is important this is not time (duration) cased as any pause is then a kill. It’s not easy as there needs to be sync, but it has to be relative, not absolute. By this, we mean it has to be relative to others and not dependant on ones own view.
Can you expand a bit on this pause? At first I thought you meant the network would appear to pause, but further comments suggest otherwise.
No I just meant the network would pause
Thank you!
It’s a pretty nice approach. Seems to me that an analogous method might work well for farming rewards too.
Yes, the trick is not to vote nodes offline (I love these types of solutions), but below a responsiveness threshold. How that works is nodes check neighbours requests/responses and if a node is less than 50% responsive relative to neighbours then we vote it is not responsive enough. There is a start position of a few messages to allow this to happen, so we don’t count the first X messages.
This is in the code now, but will definitely get tweaked as we go to find the optimal path and hopefully with as few magic numbers as possible.
I’m sure our little community would love to see a link to that interview when/if available?
How much work left in DBC?