Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Forge test for testing throughput on network purely tuned for throughput #8595

Merged
merged 1 commit into from
Jun 13, 2023
+67 −7
Merged

Forge test for testing throughput on network purely tuned for throughput #8595

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions config/src/config/consensus_config.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,8 +102,8 @@ impl Default for ConsensusConfig {
max_sending_block_bytes: 600 * 1024, // 600 KB
max_sending_block_bytes_quorum_store_override: 5 * 1024 * 1024, // 5MB
max_receiving_block_txns: 10000,
max_receiving_block_txns_quorum_store_override: 2
* MAX_SENDING_BLOCK_TXNS_QUORUM_STORE_OVERRIDE,
max_receiving_block_txns_quorum_store_override: 10000
.max(2 * MAX_SENDING_BLOCK_TXNS_QUORUM_STORE_OVERRIDE),
max_receiving_block_bytes: 3 * 1024 * 1024, // 3MB
max_receiving_block_bytes_quorum_store_override: 6 * 1024 * 1024, // 6MB
max_pruned_blocks_in_mem: 100,
Expand Down
70 changes: 65 additions & 5 deletions testsuite/forge-cli/src/main.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@
// SPDX-License-Identifier: Apache-2.0

use anyhow::{format_err, Context, Result};
use aptos_config::config::ConsensusConfig;
use aptos_config::config::{ChainHealthBackoffValues, ConsensusConfig, PipelineBackpressureValues};
use aptos_forge::{
args::TransactionTypeArg,
success_criteria::{LatencyType, StateProgressThreshold, SuccessCriteria},
Expand Down Expand Up @@ -488,11 +488,12 @@ fn single_test_suite(test_name: &str, duration: Duration) -> Result<ForgeConfig>
let single_test_suite = match test_name {
// Land-blocking tests to be run on every PR:
"land_blocking" => land_blocking_test_suite(duration), // to remove land_blocking, superseeded by the below
"realistic_env_max_throughput" => realistic_env_max_throughput_test_suite(duration),
"realistic_env_max_throughput" => realistic_env_max_throughput_test(duration),
"compat" => compat(),
"framework_upgrade" => upgrade(),
// Rest of the tests:
"realistic_env_load_sweep" => realistic_env_load_sweep_test(),
"realistic_network_tuned_for_throughput" => realistic_network_tuned_for_throughput_test(),
"epoch_changer_performance" => epoch_changer_performance(),
"state_sync_perf_fullnodes_apply_outputs" => state_sync_perf_fullnodes_apply_outputs(),
"state_sync_perf_fullnodes_execute_transactions" => {
Expand Down Expand Up @@ -910,7 +911,7 @@ fn graceful_overload() -> ForgeConfig {
ForgeConfig::default()
.with_initial_validator_count(NonZeroUsize::new(10).unwrap())
// if we have full nodes for subset of validators, TPS drops.
// Validators without VFN are proposing almost empty blocks,
// Validators without VFN are not creating batches,
// as no useful transaction reach their mempool.
// something to potentially improve upon.
// So having VFNs for all validators
Expand Down Expand Up @@ -957,7 +958,7 @@ fn three_region_sim_graceful_overload() -> ForgeConfig {
ForgeConfig::default()
.with_initial_validator_count(NonZeroUsize::new(20).unwrap())
// if we have full nodes for subset of validators, TPS drops.
// Validators without VFN are proposing almost empty blocks,
// Validators without VFN are not creating batches,
// as no useful transaction reach their mempool.
// something to potentially improve upon.
// So having VFNs for all validators
Expand Down Expand Up @@ -1367,7 +1368,7 @@ fn land_blocking_test_suite(duration: Duration) -> ForgeConfig {
}

// TODO: Replace land_blocking when performance reaches on par with current land_blocking
fn realistic_env_max_throughput_test_suite(duration: Duration) -> ForgeConfig {
fn realistic_env_max_throughput_test(duration: Duration) -> ForgeConfig {
ForgeConfig::default()
.with_initial_validator_count(NonZeroUsize::new(20).unwrap())
.with_initial_fullnode_count(10)
Expand Down Expand Up @@ -1420,6 +1421,65 @@ fn realistic_env_max_throughput_test_suite(duration: Duration) -> ForgeConfig {
)
}

fn realistic_network_tuned_for_throughput_test() -> ForgeConfig {
ForgeConfig::default()
.with_initial_validator_count(NonZeroUsize::new(12).unwrap())
// if we have full nodes for subset of validators, TPS drops.
// Validators without VFN are not creating batches,
// as no useful transaction reach their mempool.
// something to potentially improve upon.
// So having VFNs for all validators
.with_initial_fullnode_count(12)
.add_network_test(MultiRegionNetworkEmulationTest {
override_config: None,
})
.with_emit_job(EmitJobRequest::default().mode(EmitJobMode::MaxLoad {
mempool_backlog: 150000,
}))
.with_node_helm_config_fn(Arc::new(move |helm_values| {
helm_values["validator"]["config"]["consensus"]
["max_sending_block_txns_quorum_store_override"] = 10000.into();
helm_values["validator"]["config"]["consensus"]["pipeline_backpressure"] =
serde_yaml::to_value(Vec::<PipelineBackpressureValues>::new()).unwrap();
helm_values["validator"]["config"]["consensus"]["chain_health_backoff"] =
serde_yaml::to_value(Vec::<ChainHealthBackoffValues>::new()).unwrap();

helm_values["validator"]["config"]["consensus"]
["wait_for_full_blocks_above_recent_fill_threshold"] = (0.8).into();
helm_values["validator"]["config"]["consensus"]
["wait_for_full_blocks_above_pending_blocks"] = 8.into();

helm_values["validator"]["config"]["consensus"]["quorum_store"]["back_pressure"]
["backlog_txn_limit_count"] = 100000.into();
helm_values["validator"]["config"]["consensus"]["quorum_store"]["back_pressure"]
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Did you have a chance to play with this number? I actually think a higher value might not be better for throughput. It likely just affects latency on low load.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

haven't, just know I used it before :)

["backlog_per_validator_batch_limit_count"] = 10.into();

helm_values["validator"]["config"]["consensus"]["quorum_store"]["back_pressure"]
["dynamic_max_txn_per_s"] = 6000.into();

// Experimental storage optimizations
helm_values["validator"]["config"]["storage"]["rocksdb_configs"]["use_state_kv_db"] =
true.into();
helm_values["validator"]["config"]["storage"]["rocksdb_configs"]
["use_sharded_state_merkle_db"] = true.into();
}))
.with_success_criteria(
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems like a lot of success criteria can be static?

Not in this diff but it would be great to be able to configure this outside of code

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we can potentially have default success criteria, that we adjust for each test if needed.

SuccessCriteria::new(8000)
.add_no_restarts()
.add_wait_for_catchup_s(60)
.add_system_metrics_threshold(SystemMetricsThreshold::new(
// Check that we don't use more than 12 CPU cores for 30% of the time.
MetricsThreshold::new(12, 30),
// Check that we don't use more than 10 GB of memory for 30% of the time.
MetricsThreshold::new(10 * 1024 * 1024 * 1024, 30),
))
.add_chain_progress(StateProgressThreshold {
max_no_progress_secs: 10.0,
max_round_gap: 4,
}),
)
}

fn pre_release_suite() -> ForgeConfig {
ForgeConfig::default()
.with_initial_validator_count(NonZeroUsize::new(30).unwrap())
Expand Down