Closed kkonevets closed 2 years ago
b07f2b3 does not fix the problem, it still makes an RPC call that is separate from the GetMultipleAccounts
call. We have to get block hash from that GetMultipleAccounts
call, either through a sysvar, or from elsewhere in the response.
We can get the slot number from GetMultipleAccounts
(both from the sysvar and in the context
part of the response). Maybe what we can do to get the block hash is to fetch the block at that slot? This also performs a separate RPC call, but because there is a unique block hash for every (finalized) slot, the output there is not a moving target, so that is fine to do in a separate request.
Good point. Fixed. Please approve the CI build.
The integration test failed :confused:
Command '('target/debug/solido', '--cluster', 'http://127.0.0.1:8899/', '--output', 'json', 'multisig', 'create-multisig', '--multisig-program-id', 'AM83tkD6iukH4LtkCR1TbP9Xp1xUJh2kCRriYR4vKvaH', '--threshold', '2', '--owners', 'GjhPQBR1wA7H9efpbFeXsRJZVkroLsUNg8Wve2fheePG,7F1ET9qn9Sm2ULkxXSfaYa99X8w3B9ufpdKPdrNDCSUV')' returned non-zero exit status 1.
We’ve seen some flakes in the integration tests in the past but it would usually fail earlier, can you check what’s up?
RPC should be done with commitment "confirmed", otherwise we get "block not available for slot ..." error
Can we pull to main now?
By the way, should we increment a version number of solido to 1.3.3 ?
We could, if this is the last change we make before the larger changes, it would make sense.
Soon I will edit this PR https://github.com/ChorusOne/solido/pull/569. So we should probably wait
Actually, we should probably update the maintainer daemon to remove the Anker steps again and then make a new release, before the larger "2.0" update.
util::ensure_orca_program_exists()
once so that SOlana does not block us by IP when running lots of testsget_latest_blockhash
instead ofRecentBlockhashes
get_new_latest_blockhash
in a normal way