Witness_node randomly stops syncing

[abitmore]

Bug Description

Start a new witness_node instance and wait, sometimes it hangs during syncing (unable to sync to latest block). Restarting works sometimes.

See https://github.com/bitshares/bitshares-core/issues/2798#issuecomment-1855022095 for more info.

Unable to stably reproduce so far. Haven't found very interesting info in log files yet. Probably caused by memory corruption.

Impacts Describe which portion(s) of BitShares Core may be impacted by this bug. Please tick at least one box.

• [ ] API (the application programming interface)
• [ ] Build (the build process or something prior to compiled code)
• [ ] CLI (the command line wallet)
• [ ] Deployment (the deployment process after building such as Docker, Travis, etc.)
• [ ] DEX (the Decentralized EXchange, market engine, etc.)
• [x] P2P (the peer-to-peer network for transaction/block propagation)
• [x] Performance (system or user efficiency, etc.)
• [ ] Protocol (the blockchain logic, consensus, validation, etc.)
• [x] Security (the security of system or user data, etc.)
• [x] UX (the User Experience)
• [ ] Other (please add below)

Host Environment Please provide details about the host environment. Much of this information can be found running: witness_node --version.

• Host OS: Ubuntu (various versions)
• Host Physical RAM Sufficient
• BitShares Version: 7.0.1 / test-7.0.3
• OpenSSL Version: -
• Boost Version: -

CORE TEAM TASK LIST

• [ ] Evaluate / Prioritize Bug Report
• [ ] Refine User Stories / Requirements
• [ ] Define Test Cases
• [ ] Design / Develop Solution
• [ ] Perform QA/Testing
• [ ] Update Documentation

[abitmore]

Found the reason.

In PR https://github.com/bitshares/bitshares-core/pull/2791 for the last releases (7.0.1 and test-7.0.3), we updated read_write_handler and read_write_handler_with_buffer to throw canceled_exception when a boost::asio::error::operation_aborted error occurs.

The canceled_exception might then be caught in message_oriented_connection_impl::read_loop(): https://github.com/bitshares/bitshares-core/blob/0aa383154ce191d708f7d143b27662bc171ed239/libraries/net/message_oriented_connection.cpp#L215-L230

as a result, call_on_connection_closed is no longer set to true, so node::on_connection_closed() is not called as before, which means that certain cleanup steps are not performed.