Use better logs (#452)

crates/common/src/http.rs

@@ -3,7 +3,7 @@ use crate::{
     stats::{Gauge, IncrementingMetric},
 };
 use derivative::Derivative;
-use log::trace;
+use log::debug;
 use proxy_wasm::traits::Context;
 use serde::Serialize;
 use std::{cell::RefCell, collections::HashMap, fmt::Debug, time::Duration};
@@ -48,7 +48,7 @@ pub trait Client: Context {
         call_args: CallArgs,
         call_context: Self::CallContext,
     ) -> Result<u32, ClientError> {
-        trace!(
+        debug!(
             "dispatching http call with args={:?} context={:?}",
             call_args,
             call_context

crates/common/src/ratelimit.rs

@@ -1,7 +1,7 @@
 use crate::configuration;
 use configuration::{Limit, Ratelimit, TimeUnit};
 use governor::{DefaultKeyedRateLimiter, InsufficientCapacity, Quota};
-use log::trace;
+use log::debug;
 use std::fmt::Display;
 use std::num::{NonZero, NonZeroU32};
 use std::sync::RwLock;
@@ -99,7 +99,7 @@ impl RatelimitMap {
         selector: Header,
         tokens_used: NonZeroU32,
     ) -> Result<(), Error> {
-        trace!(
+        debug!(
             "Checking limit for provider={}, with selector={:?}, consuming tokens={:?}",
             provider,
             selector,

crates/common/src/tokenizer.rs

@@ -1,14 +1,14 @@
-use log::trace;
+use log::{debug};
 
 #[allow(dead_code)]
 pub fn token_count(model_name: &str, text: &str) -> Result<usize, String> {
-    trace!("getting token count model={}", model_name);
+    debug!("getting token count model={}", model_name);
     //HACK: add support for tokenizing mistral and other models
     //filed issue https://github.com/katanemo/arch/issues/222
 
     let updated_model = match model_name.starts_with("gpt") {
         false => {
-            trace!(
+            debug!(
                 "tiktoken_rs: unsupported model: {}, using gpt-4 to compute token count",
                 model_name
             );