rpc task loop when umount an NFS 4.x share with sec=krb5 triggered by systemd / autofs
Issue
- High CPU usage in
sys
area when autofs/systemd umount an NFS4.x share withsec=krb5
. - Time is consumed by
rpciod
kworker task, spawned when umount is triggered. - A reboot is needed to recover from this situation.
Environment
- Red Hat Enterprise Linux 8.3 or higher
- 8.3 kernels (4.18.0-240.*el8) before kernel-4.18.0-240.10.1.el8_3
- NFS 4.x share with
sec=krb5
- autofs/systemd
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.