Log a warning when scheduled job args are over 1MB (#42379)

We would like to put a limit on these; as a first step let's see how common it is.

GitOrigin-RevId: e5936f68cad6e8304e7722bf55b28cede9475d14

Author: goffrie

crates/common/src/knobs.rs

@@ -22,6 +22,7 @@ use std::{
 };
 
 use cmd_util::env::env_config;
+use value::MAX_USER_SIZE;
 
 use crate::fastrace_helpers::SamplingConfig;
 
@@ -246,7 +247,14 @@ pub static TRANSACTION_MAX_NUM_SCHEDULED: LazyLock<usize> =
 pub static MAX_JOBS_CANCEL_BATCH: LazyLock<usize> =
     LazyLock::new(|| env_config("MAX_JOBS_CANCEL_BATCH", 1000));
 
-/// Maximum size of the arguments to a scheduled function.
+/// Maximum size of a single scheduled function's arguments.
+/// This is not currently enforced.
+pub static MAX_SCHEDULED_JOB_ARGUMENT_SIZE_BYTES: LazyLock<usize> = LazyLock::new(|| {
+    env_config("MAX_SCHEDULED_JOB_ARGUMENT_SIZE_BYTES", MAX_USER_SIZE) // 1 MiB
+});
+
+/// Maximum total size of the arguments to all functions scheduled in a single
+/// transaction.
 pub static TRANSACTION_MAX_SCHEDULED_TOTAL_ARGUMENT_SIZE_BYTES: LazyLock<usize> =
     LazyLock::new(|| {
         env_config(

crates/model/src/scheduled_jobs/mod.rs

@@ -14,6 +14,7 @@ use common::{
     },
     execution_context::ExecutionContext,
     knobs::{
+        MAX_SCHEDULED_JOB_ARGUMENT_SIZE_BYTES,
         TRANSACTION_MAX_NUM_SCHEDULED,
         TRANSACTION_MAX_SCHEDULED_TOTAL_ARGUMENT_SIZE_BYTES,
     },
@@ -175,7 +176,6 @@ impl<'a, RT: Runtime> SchedulerModel<'a, RT> {
                 )
             )
         );
-        self.tx.scheduled_size.num_writes += 1;
         anyhow::ensure!(
             self.tx.scheduled_size.size + args.size()
                 <= *TRANSACTION_MAX_SCHEDULED_TOTAL_ARGUMENT_SIZE_BYTES,
@@ -188,6 +188,10 @@ impl<'a, RT: Runtime> SchedulerModel<'a, RT> {
                 )
             ),
         );
+        if args.size() > *MAX_SCHEDULED_JOB_ARGUMENT_SIZE_BYTES {
+            tracing::warn!("Scheduling a job with argument size {}", args.size());
+        }
+        self.tx.scheduled_size.num_writes += 1;
         self.tx.scheduled_size.size += args.size();
         Ok(())
     }