[1.0-beta4] Misc improvements #448
Conversation
…ocks. Also refactor terminate-at-block handling.
|
Does this mean that if 1/3 of finalizer are off-line for 10 minutes, all nodes will shutdown? |
Closer to 8.3 minutes, yes. |
This doesn't seem desirable though. Maybe it would be better to just skip more and more block producing slots until we start getting QCs again? For example every 10 minutes, skip twice as many producing slots. |
| @@ -4701,6 +4702,25 @@ struct controller_impl { | |||
| return conf.block_validation_mode == validation_mode::LIGHT || conf.trusted_producers.count(producer); | |||
| } | |||
|
|
|||
| bool should_terminate() const { | |||
| constexpr block_num_type max_reversible_blocks = 1000; | |||
There was a problem hiding this comment.
Temporary solution until we can get the consensus level change that will restrict producing higher block heights until finalizers come back online.
There was a problem hiding this comment.
We have some libtester tests that failed because they produce more than 1000 blocks (more than 2300 if I counted correctly). How about we go with 30 minutes; 3600 blocks.
There was a problem hiding this comment.
As long as we have some protection against OOM, 30 minutes is fine for now.
see https://github.com/AntelopeIO/spring/pull/448/files#r1697549049 |
heifner
changed the title
|
Note:start |
Github auto-merged this because #451 was built off the same branch. Ignore this PR.