blockchain on disk is saved equally on all nodes (no binary difference in any block found)
what we do:
- start new testnet and check if this issue can occur even if no transactions ever been made (up to some days)
- investigate error logs on all nodes
- investigate program code thoroughly again
Last edit: exxoo 2014-06-15
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
it cannot be that random. Well it can happen at any block id but it always happens on all connected peers and on all server in the core net at the same time (at least at the same block).
So we can forget to think about thread concurrency/race conditions.
Also, the CUdp class and the udp messages are ok and cannot be the reason for this issue.
That brings the focus back to the classes CChain, CBlock, CTransaction and the sha256 function.
More information about this problem in ~2-4 days.
With the latest test in the testnets (especially the test to let the network run without any transactions made) will show if this is somehow linked to initiated transactions at all.
Last edit: exxoo 2014-06-18
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
If this urgent, major bug is not solved after the weekend we will prepare a full statement (description, relevant code parts partially, what we tried and what we already found out, errorLogs of server) and will ask you, if you like, to have a look at the issue.
We hope to find the cause as soon as possible.
Last edit: exxoo 2014-06-19
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
but thats not the full solution. The full(hopefully!) solution is working now in theory and will be programmed as soon as possible (1-3 days with testing).
More information on this issue soon.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
what we do:
- start new testnet and check if this issue can occur even if no transactions ever been made (up to some days)
- investigate error logs on all nodes
- investigate program code thoroughly again
Last edit: exxoo 2014-06-15
Diff:
it cannot be that random. Well it can happen at any block id but it always happens on all connected peers and on all server in the core net at the same time (at least at the same block).
So we can forget to think about thread concurrency/race conditions.
Also, the CUdp class and the udp messages are ok and cannot be the reason for this issue.
That brings the focus back to the classes CChain, CBlock, CTransaction and the sha256 function.
More information about this problem in ~2-4 days.
With the latest test in the testnets (especially the test to let the network run without any transactions made) will show if this is somehow linked to initiated transactions at all.
Last edit: exxoo 2014-06-18
If this urgent, major bug is not solved after the weekend we will prepare a full statement (description, relevant code parts partially, what we tried and what we already found out, errorLogs of server) and will ask you, if you like, to have a look at the issue.
We hope to find the cause as soon as possible.
Last edit: exxoo 2014-06-19
we made progress on this issue. It is partially linked to the (new created) ticket https://sourceforge.net/p/exocoin/tickets/31/
but thats not the full solution. The full(hopefully!) solution is working now in theory and will be programmed as soon as possible (1-3 days with testing).
More information on this issue soon.
the solution is implemented to about 75%. Including testing we expect to have a fully working solution in ~2 days.
the solution not only should solve this issue but also enhance the security against malicious attacks on the network.
v0.61 is almost done, especially this issue.
Testing will need some time.
building up more testcases to verify the mechanisms since they got quite complex.
all suceeded except one. Currently working on this.
re-running test. Should be ok now
seems to work! To be 100% sure we would need the next beta test and see what happens.
Next beta test will be started soon...
we consider this as "fixed" unless 0.61 will show we are wrong.