1. Deploy `some long running code` and make node propose it, e.g.
rnode deploy --private-key=bd2946a79be97625b86c09510a48aa037905c408a73fc672f19068dc58c550ee --phlo-limit 10000000000000 --phlo-price 1 infinite-sends-and-receives.rho
2. Ctrl+C node
3. Restart node.
4. As block was not created on step 1, I expect node trying to include `some long running code` in the next block, so never succeeding in creating new block. But if you try deploy anything else after node restart and propose - node successfully creates block.
Which is a feature - because infinite loops does not stuck in mempool. But also is a bug, because node can drop legit deploys if it restarts in the middle of theirs execution.
Taking into account that validators have to have high uptime, this is a `precision asteroid` case, but anyway, looks like potential data loss scenario