fix(pipeline): clear MerkleStage checkpoints on invalid root#3973
fix(pipeline): clear MerkleStage checkpoints on invalid root#3973
MerkleStage checkpoints on invalid root#3973Conversation
joshieDo
added
C-bug
joshieDo
changed the title
MerkleStage checkpoints on invalid root
Codecov Report
... and 10 files with indirect coverage changes
Flags with carried forward coverage won't be shown. Click here to find out more.
|
rkrasiuk
left a comment
rkrasiuk
left a comment
There was a problem hiding this comment.
nice catch! however, we should do this inside the merkle stage
|
we cannot commit anything from inside a stage though, and I do like that constraint furthermore, there's already a branch to handle the validation error, so as long as it doesn't grow with many edge cases I feel like it should be fine? edit: and we cannot drop the current transaction either |
|
ah, ye, makes sense |
onbjerg
left a comment
onbjerg
left a comment
There was a problem hiding this comment.
Would like to mark this as a FIXME since it's technically debt we should try to solve in a better way (this is "special treatment" for merkle)
onbjerg
force-pushed
the
merkle/clean-checkpoint
branch
from
ea90b5e to
6b339a3
Compare
3 participants
While working on another branch, I noticed that the merkle stage once it fails its validation (related to my branch), it can't really recover since its checkpoint never gets cleared.
This is what happened before the logs from the image below.
AccountHashingandStorageHashing.AccountHashingStorageHashing(you can see its end in the first line of the picture.Proposing that we clear
MerkleStagecheckpoints once we hit a validation error.