You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Nonces stored in the --user-data-dir might get out of sync with the network for various reasons. To avoid failing workloads due to invalid nonces, they can be queried from the RPC. This approach works well when there are not too many accounts in --user-data-dir.
Possible steps for implementing this:
Here is some code that get's access keys for a set of accounts. Factor that out to make it reusable in multiple places. Ideally also handling the TODO to use a JoinSet.
Once above is done, use that code when starting up workloads like benchmark-native-transfers or benchmark-mpc-sign to query recent nonces.
It should be possible to enable/disable this with a CLI flap. In a scenario with a lot of accounts and knowing that their nonces on disk are fine, we want to avoid querying nonces from the RPC.
With this in place, is #12753 still relevant? I would argue yes, because #12753 is a relatively small change and will help in the scenario with a lot of accounts.
The text was updated successfully, but these errors were encountered:
[issue 12805](#12805)
The flag (default=false) triggers the update of the account nonces from the network before generating the transactions.
impl details:
- extract the `update_account_nonces()` function
- add a corresponding flag to benhmark_native_transfers and benchmark_mpc binaries
+ apply the `update_account_nonces()` depending on the flag value
tested:
- run the benchmark from the clean state
- edit the `a_user_0.test.near.json`, set "nonce" to 7000001
- update the log level to "trace" in the justfile
- re-run the benchmark
- find the `[2025-01-29T13:06:56Z TRACE near_synth_bm::account] updating nonce for a user a_user_0.test.near (7000001->7000020)` in the log
Nonces stored in the
--user-data-dir
might get out of sync with the network for various reasons. To avoid failing workloads due to invalid nonces, they can be queried from the RPC. This approach works well when there are not too many accounts in--user-data-dir
.Possible steps for implementing this:
JoinSet
.benchmark-native-transfers
orbenchmark-mpc-sign
to query recent nonces.With this in place, is #12753 still relevant? I would argue yes, because #12753 is a relatively small change and will help in the scenario with a lot of accounts.
The text was updated successfully, but these errors were encountered: