feat(cli): adds load and state cli options #502
Conversation
dutterbutter
added
the
do not merge 🛑
dutterbutter
changed the title
dutterbutter
removed
the
do not merge 🛑
dutterbutter
force-pushed
the
db/adds-load-state-cli-option
branch
from
d62332d to
9c0e73d
Compare
|
|
||
| // Allow some time for the state to be dumped | ||
| sleep(Duration::from_secs(2)); |
There was a problem hiding this comment.
Suggest keeping sleep here, as sometimes the json file is not dumped in time.
There was a problem hiding this comment.
I think we need to add a sigterm handler to enforce dumping. I will do some experiments later
| let new_provider = init_testing_provider(move |node| { | ||
| node.path(get_node_binary_path()) | ||
| .arg("--state-interval") | ||
| .arg("1") | ||
| .arg("--load-state") | ||
| .arg(dump_path_clone.to_str().unwrap()) | ||
| .fork("sepolia-testnet") | ||
| }) | ||
| .await?; |
There was a problem hiding this comment.
@itegulov I've noticed some flakiness with the fork tests timing out during initialization (e.g. test has been running for longer then x).
Any recommendations to harden?
There was a problem hiding this comment.
Yeah I had a feeling this would be an issue. I think we should not do fork tests in this form right now. To minimize flakiness we should set up a real local ZKsync enviornment and use as forking source in all of the fork e2e tests. For example we could spin up dockerized zksync-era node and then seal some blocks.
In any case I think it's for the better to delete these for now.
There was a problem hiding this comment.
Skipping the fork tests for now, and created issue: #508
| @@ -548,6 +548,7 @@ impl ForkDetails { | |||
| ) | |||
| })?; | |||
| let l1_batch_number = block_details.l1_batch_number; | |||
| block.l1_batch_number = Some(l1_batch_number.0.into()); | |||
There was a problem hiding this comment.
TBH I have feeling this might lead to some unexpected behaviour because that block might technically not be sealed in that batch yet (this is what l1BatchNumber == null means). But anyway, the whole forking logic needs some rethinking in general so I am fine with keeping as is
…s/anvil-zksync into db/adds-load-state-cli-option
What 💻
--load-stateand--statecmds to cli"l1BatchNumber": null,issueWhy ✋
Evidence 📷
Include screenshots, screen recordings, or
consoleoutput here demonstrating that your changes work as intendedRun the following command to first easily create a state.json file:
Then we can use that state to start the node using
--load-stateForking:
Then we can use that state to start the node using
--load-state