efs-utils making the stunnel enter zombie state.

[Sharukh95]

We're facing an issue where the stunnel process running on our AWS Ec2 enters a zombie state. This results in the nfs server timeout (We're using EFS). We're running Amazon Linux 2 (Kernel version 4.14.318-241.531) with stunnel version 5.6.4. The EFS utils version we're using is 3.1.33

[paurosello]

We are facing the same issue on kubernetes, when the CSI driver reboots, connection to the EFS drive timeouts

[Ashley-wenyizha]

Hi, thanks for reporting this issue, assuming you are using 1.33.3? The EFS utils version we're using is 3.1.33

We fixed the zombie stunnel issue in a later version, could you upgrade to latest version and see if it still persists?

thanks

[whites11]

Hi, thanks for reporting this issue, assuming you are using 1.33.3? The EFS utils version we're using is 3.1.33

We fixed the zombie stunnel issue in a later version, could you upgrade to latest version and see if it still persists?

thanks

What version exactly was it fixed?
We're still seeing this in 1.35.0:

# ./amazon-efs-mount-watchdog --version
./amazon-efs-mount-watchdog Version: 1.35.0

(I'm @paurosello's teammate)

[luca-rui]

@Ashley-wenyizha Has there been any news on this topic?

(also a teammate of @paurosello and @whites11 😅)

[phmeier-nubank]

I have observed zombie processes as well with 1.35.0. This is a patch which fixed it for me in a test system.

From 87c1f0169a003e7ef5a0297e1d7aaaacdb19e91b Mon Sep 17 00:00:00 2001
From: Philipp Meier <philipp.meier@nubank.com.br>
Date: Thu, 1 Feb 2024 16:32:07 +0100
Subject: [PATCH] Read subprocess status to prevent zombies

---
 src/watchdog/__init__.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/src/watchdog/__init__.py b/src/watchdog/__init__.py
index a419f80..05c23b0 100755
--- a/src/watchdog/__init__.py
+++ b/src/watchdog/__init__.py
@@ -1262,6 +1262,9 @@ def check_stunnel_health(
         # process after the timeout.
         #
         process.kill()
+    finally:
+        # read proc state to prevent zombies
+        process.wait()


 # Retrieve the nfs mountpoint with the port information in the mount option
@@ -1316,6 +1319,8 @@ def check_child_procs(child_procs):
                 proc.pid,
                 proc.returncode,
             )
+            # read proc state to prevent zombies
+            proc.wait()
             child_procs.remove(proc)


--
2.43.0

Use git am < this-file.path to apply.

To verify I have conducted some tests including killing and halting the stunnel process. Before the patch I could observe zombie processes.