P2P sync failing

[heifner]

This issue is related to #291 stability.

Current sync of producer nodes is not working. Steps to reproduce:

• programs/launcher/launcher -p5 -s mesh --skip-signature
  • -s star, -s ring also exhibit same issue
• programs/eosc/eosc --port 8888 get block 1
• programs/eosc/eosc --port 8889 get block 1
• programs/eosc/eosc --port 8890 get block 1
• programs/eosc/eosc --port 8891 get block 1
• programs/eosc/eosc --port 8892 get block 1
  • The blocks returned do not agree.
  • Also the following is in tn_data_*/stderr.txt

426004ms thread-0   net_plugin.cpp:820            handle_message       ] unable to accept block #3 syncing
426004ms thread-0   fork_database.cpp:65          push_block           ] Pushing block to fork database that failed to link: 00000003dcb43dac9f0083ea66e8f62ca0eaf0b6141894412ee3557c4b7a6a18, 3
426004ms thread-0   fork_database.cpp:66          push_block           ] Head: 2, 000000020f5b636cd1777ce3453373501ce6bbccd158725470d91ca855821dde

• Large amounts of memory were consumed by the eosd

[heifner]

Created a simple test and handed it over to Phil who is actively working this issue with #406. It is much more stable in branch p2p-txn-handling-393, but still can be made to fail via dropped connections.

[pmesnier]

I've created a new branch off of "staging" to carry on my work sorting out the p2p networking functionality. This issue relates back to #291 and the working branch is p2p-sync-slow-peers.

[pmesnier]

Since I'm not actually working this issue directly, I'm removing the in progress label for now

[pmesnier]

I'm pretty sure this is failing simply because the last bit of sync code isn't in place yet. This is the part involving fetching the blocks post irreversible.

[pmesnier]

Solution added with PR #543

[pmesnier]

PR #543 addresses this issue