agent: Fix an issue reporting OOM events by mistake

The agent registers an event fd in `memory.oom_control`. An OOM event is
forwarded to containerd when the event is emitted, regardless of the
content in that file.

I observed content indicating that events should not be forwarded, as shown
below. When `oom_kill` is set to 0, it means no OOM has occurred. Therefore,
it is important to check the content to avoid mistakenly forwarding OOM
events.

```
oom_kill_disable 0
under_oom 0
oom_kill 0
```

Fixes: #8715

Signed-off-by: Xuewei Niu <niuxuewei.nxw@antgroup.com>

src/agent/rustjail/src/cgroups/notifier.rs

@@ -3,19 +3,20 @@
 // SPDX-License-Identifier: Apache-2.0
 //
 
-use anyhow::{anyhow, Context, Result};
-use eventfd::{eventfd, EfdFlags};
-use nix::sys::eventfd;
 use std::fs::{self, File};
 use std::os::unix::io::{AsRawFd, FromRawFd};
 use std::path::Path;
 
-use crate::pipestream::PipeStream;
+use anyhow::{anyhow, Context, Result};
+use eventfd::{eventfd, EfdFlags};
 use futures::StreamExt as _;
 use inotify::{Inotify, WatchMask};
+use nix::sys::eventfd;
 use tokio::io::AsyncReadExt;
 use tokio::sync::mpsc::{channel, Receiver};
 
+use crate::pipestream::PipeStream;
+
 // Convenience function to obtain the scope logger.
 fn sl() -> slog::Logger {
     slog_scope::logger().new(o!("subsystem" => "cgroups_notifier"))
@@ -165,22 +166,17 @@ async fn register_memory_event(
 
     tokio::spawn(async move {
         loop {
-            let sender = sender.clone();
             let mut buf = [0u8; 8];
             match eventfd_stream.read(&mut buf).await {
                 Err(err) => {
                     warn!(sl(), "failed to read from eventfd: {:?}", err);
                     return;
                 }
                 Ok(_) => {
-                    let content = fs::read_to_string(path.clone());
-                    info!(
-                        sl(),
-                        "cgroup event for container: {}, path: {:?}, content: {:?}",
-                        &containere_id,
-                        &path,
-                        content
-                    );
+                    // Do not send an OOM event in the case where no OOM has occurred
+                    if get_value_from_cgroup(&path, "oom_kill").unwrap_or(0) <= 0 {
+                        continue;
+                    }
                 }
             }