Small package of fixes for migration process #14558
Conversation
|
ci-build-me! |
|
Did you see my message about writing the migrated transaction hashes? When migrated, the old transaction hashes will be written as new transaction hashes. In my message, I explained how to avoid that. You might want to add a check that the migrated transaction hashes are syntactically the same as the old ones. |
dkijania
force-pushed
the
dkijania/migration_app_enh
branch
from
3e5a7a3 to
388d097
Compare
| @@ -142,7 +143,7 @@ let user_commands_from_block_id ~mainnet_pool block_id = | |||
| ~f:mainnet_transaction_failure_of_string | |||
| in | |||
| let hash = | |||
| Mina_transaction.Transaction_hash.of_base58_check_exn user_cmd.hash | |||
| Mina_transaction.Transaction_hash.of_base58_check_exn_v1 user_cmd.hash | |||
There was a problem hiding this comment.
I think you should also change in archive_lib.user_commands.add_extensional_if_doesn't_exist function. Otherwise the hash you writen to the migrated database would be changed:
mina/src/app/archive/lib/processor.ml
Line 1882 in 79c6253
There was a problem hiding this comment.
I sent an email to Deepthi with a suggested way to handle this issue. You'd add an optional boolean flag to add_extensional_if_doesn't_exist that would indicate whether to use the V1 hash.
|
I agree with paul that you want to add some check that checks the migrated transaction hash is syntactically the same as before. |
dkijania
force-pushed
the
dkijania/migration_app_enh
branch
from
7e1d946 to
75222cb
Compare
|
!ci-build-me |
src/app/archive/lib/processor.ml
Outdated
| @@ -34,6 +34,13 @@ let applied_str = "applied" | |||
|
|
|||
| let failed_str = "failed" | |||
|
|
|||
| let to_base58_check ?(v1_transaction_hash = false) hash = | |||
| match v1_transaction_hash with | |||
There was a problem hiding this comment.
if ... then ... else might be simpler than match here
|
!ci-build-me |
|
!ci-nightly-me |
|
!approved-for-mainnet |
Archive Migration tooling consist of two applications:
During testing I found a couple of improvements/ small fixes which are necessary for testing and correct run:
a) bucket location
b) number of batches for precomputed block fetch
c) corner case when there are no precomputed blocks to fetch
Closes #14551