Skip to content

Conversation

@emberian
Copy link
Contributor

No description provided.

mitschabaude and others added 30 commits March 8, 2024 15:16
Retry fork config export on empty or null response
Problem: fork config export requires stop-tx-slot to be at depth less
than `k`.

Solution: bump `k` and shorten window between stop-tx-slot and
stop-chain-slot.
Plus fix checking emptiness check for blocks created after
slot-chain-end
This commit moves script files and introduces a script that builds
necessary dependencies and can be used both locally and from CI.

Additionally, a CI job is introduced to executed the HF unit test as
part of nightlies.
Nix image that is distributed on dockerhub doesn't have the latest
nixpkgs. In particular it uses nixpkgs that doesn't fully support
`?submodules=1` in flakes (case of submodules having submodules of their
own, which triggered certain issues when building `compatible` branch).
When CI provides the necessary secret env variables (or they are
provided locally), cache is updated with built binaries.

This makes sure that repretetive runs do not need to rebuild binaries.
But the most useful use case is that if `compatible` branch is updated
rarely, it will not be rebuilt between updates, reducing the runtime of
test by a few minutes at least.
This ensures that the test is randomized between runs, while allowing
CI or user to specify the value that needs to be tested.
Implement checks for all ledger hashes that are used for generation of
runtime config for the fork network.

Solution continuously queries local network to capture last block of
every epoch. These blocks contain snarked ledger hashes that will later
(with the delay of 2 epochs) be used as staking/next ledgers.

Then fork config is checked to contain the correct seeds (which are also
in scope of this PR).

This PR contains local patch against compatible to
port runtime_genesis_ledger from berkeley (without sha3 hashing). This
utility is to be executed after pre-fork network is stopped to generate
hashes file using hashes of `compatible`. Then resulting hashes are
compared against prevously captured snarked ledgers (in edge case,
staking/next epoch ledgers of genesis epoch). This ensures fork config
export (that is executed after slot_chain_end) reports correct ledgers.
This reverts commit cf4f6c4.
deepthiskumar and others added 26 commits March 27, 2024 14:18
…rsion' into dkijania/separate_berkeley_migration_package
…we need to determine toolchain not based on profile. I introduce custom toolchain selection
…ission-for-verification-key-version

make replayer --migration-mode to overwrite the versions
…ompatible' into dkijania/port_guardian_script_to_berk
…ript_to_berk

Port missing-blocks-guardian and include it into deb pkg to berk
…nance_docker_berk

Port mina maintenance docker berk
…y_migration_package

Reverting CI code to build sperate debian exclusively for berkeley migration package
@emberian emberian requested review from a team as code owners March 28, 2024 22:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

9 participants