Fix fee catchup after restart #2160
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jbearer
requested review from
philippecamacho,
ImJeremyHe,
sveitser,
tbro and
imabdulbasit
as code owners
This test reproduces an issue discovered in staging, in which, after a staggered restart of all nodes, the network has forgotten the fee state for all accounts except those which actively produced blocks during the restart. Since the state is needed for undecided blocks, it cannot directly be obtained from persisted merklized state, which is only available for decided blocks.
jbearer
force-pushed
the
jb/state-catchup-2
branch
from
79dc84c to
ef41777
Compare
jbearer
force-pushed
the
jb/state-catchup-2
branch
from
79d17b9 to
d56be55
Compare
…e HotShot state map
bfish713
reviewed
Oct 11, 2024
During experimentation I changed the restart tests from starting up all nodes in parallel to one at a time. This broke the no-CDN tests because, in some cases, you need multiple nodes to restart before libp2p can become ready, but this blocks the initialization of the first restarted node. Switched it back to starting up in parallel and things are working again.
|
@bfish713 can you also look at the state map update during catchup? (This basically acts as a cache for successful catchup requests) |
sveitser
reviewed
Oct 11, 2024
Co-authored-by: Mathis <[email protected]>
jbearer
added a commit
that referenced
this pull request
Oct 14, 2024
This is a backport of #2160 to release-gambit. It is a separate PR because it is not a clean cherry-pick. I recommend reviewing the `main` PR first, and then you can just look at this one to convince yourself it's doing the same thing.
Took a look and that all looks fine to me |
The simplest, most robust change was to make the task a flat retry loop, instead of having a nested retry loop for the fetch itself, and to update the anchor view each iteration of the loop, so that the loop exits if we reach a decide and the anchor view becomes more recent than the view we are fetching.
* Enable frontier catchup from storage * Prevent recursive state reconstruction * Prefetch accounts used for paying fees during state reconstruction
bfish713
approved these changes
Oct 17, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR reproduces and will fix an issue discovered in staging, in which, after a staggered restart of all nodes, the network has forgotten the fee state for all accounts except those which actively produced blocks during the restart. Since the state is needed for undecided blocks, it cannot directly be obtained from persisted merklized state, which is only available for decided blocks.
This PR:
CatchupStoragetrait, distinct fromCatchupDataSource. This allows us to get more information when querying storage, such as the leaf corresponding to fetched account state, which in turn ensures we are always able to cache fetched state in the HotShot view mapKey places to review