banks client CU meter not working for solana-program-test

[cavemanloverboy]

For no good reason, I have a suspicion this is a MacOS problem. I don't have a linux machine to test on, but the folks I've been in contact with say this works on their linux machine.

If I take the example-helloworld program and insert this diff

% git diff tests
diff --git a/src/program-rust/tests/lib.rs b/src/program-rust/tests/lib.rs
index cc9cc02..890b43e 100644
--- a/src/program-rust/tests/lib.rs
+++ b/src/program-rust/tests/lib.rs
@@ -54,7 +54,8 @@ async fn test_helloworld() {
         Some(&payer.pubkey()),
     );
     transaction.sign(&[&payer], recent_blockhash);
-    banks_client.process_transaction(transaction).await.unwrap();
+    let meta = banks_client.process_transaction_with_metadata(transaction).await.unwrap();
+    println!("CUs: {}", meta.metadata.unwrap().compute_units_consumed);

     // Verify account has one greeting
     let greeted_account = banks_client

to print CUs, I always get:

running 1 test
[2023-12-16T19:02:55.406071000Z INFO  solana_program_test] "helloworld" builtin program
[2023-12-16T19:02:55.429180000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM invoke [1]
[2023-12-16T19:02:55.429236000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM invoke [1]
[2023-12-16T19:02:55.429320000Z DEBUG solana_runtime::message_processor::stable_log] Program log: Hello World Rust program entrypoint
[2023-12-16T19:02:55.429328000Z DEBUG solana_runtime::message_processor::stable_log] Program log: Greeted 1 time(s)!
[2023-12-16T19:02:55.429335000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM success
[2023-12-16T19:02:55.429369000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM success
CUs: 0
[2023-12-16T19:02:55.430223000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM invoke [1]
[2023-12-16T19:02:55.430247000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM invoke [1]
[2023-12-16T19:02:55.430315000Z DEBUG solana_runtime::message_processor::stable_log] Program log: Hello World Rust program entrypoint
[2023-12-16T19:02:55.430324000Z DEBUG solana_runtime::message_processor::stable_log] Program log: Greeted 2 time(s)!
[2023-12-16T19:02:55.430328000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM success
[2023-12-16T19:02:55.430349000Z DEBUG solana_runtime::message_processor::stable_log] Program 1111111QLbz7JHiBTspS962RLKV8GndWFwiEaqKM success
test test_helloworld ... ok

Not only is the reported CUs zero, it appears I can essentially do infinite compute within the instruction. As such, the test environment is not ideal.

[ifiokjr]

@cavemanloverboy was this ever resolved? Is that why you closed it?

I'm seeing very low CU values in my BanksClient tests. I'm wondering how to check the actual CU without running a local validator.

[cavemanloverboy]

no i just learned that it's not actually metered (on this version at least) when not running the bpf program