On RHCOS worker nodes in OpenShift, mounting an NFS directory to /var/lib/systemd/coredump (via bind mount) causes systemd-coredump to stop generating new core dump files.
The same configuration works correctly on standard RHEL nodes.
Core dumps are successfully generated before the NFS bind mount. After stopping systemd-coredump.socket and performing a bind mount of an NFS directory onto /var/lib/systemd/coredump, no new core dumps are generated and coredumpctl list shows nothing.
This behavior appears specific to RHCOS.    

    100% reproducible on RHCOS worker nodes.

   1️⃣ Generate core dump before NFS mount
From test pod:
oc rsh <httpd-pod>
kill -SIGABRT <pid>

On node:
oc debug node/<worker>
chroot /host
ls -lh /var/lib/systemd/coredump/

Result:

Core dump files are generated successfully.

    2️⃣ Mount NFS on node
Inside oc debug → chroot /host:
Verify NFS:
ping -c 2 <nfs-server>
showmount -e <nfs-server>

Create mount point:
mkdir -p /var/mnt/nfs-coredumps

Mount NFS manually:
mount -t nfs -o vers=4 <nfs-server>:/nfs/coredumps /var/mnt/nfs-coredumps

Verify:
df -h | grep coredumps
touch /var/mnt/nfs-coredumps/test-from-node

3️⃣ Stop socket and bind mount
systemctl stop systemd-coredump.socket
mount --bind /var/mnt/nfs-coredumps /var/lib/systemd/coredump
mount | grep coredump

4️⃣ Generate new core dump
From test pod:
kill -SIGABRT <pid>

Check on node:
ls -lh /var/lib/systemd/coredump/
coredumpctl list

    No new core files generated


coredumpctl list shows nothing


No files written to NFS mount


systemd-coredump appears to silently fail

    Core dumps should be written to the NFS-backed directory


coredumpctl list should display new entries


Behavior should match RHEL behavior

    this seems very specific to systemd