RHEL 7.6: NFS share hangs when NFS operations are in-progress and a network partition occurs involving an unexpected RST
Issue
- After upgrading from the RHEL7.5 kernel (3.10.0-862*el7) to the RHEL7.6 kernel (3.10.0-957*el7), reads or writes to a NFS share by a NFS client hang indefinitely following a network partition and recovery.
- NFS share is hung with a task, such as
kworker
, consuming 100% of CPU and the/proc/<pid>/stack
or other backtrace shows execution in the sunrpc transmit path - Processes can not be killed and a reboot of the NFS client is required.
- This is new behaviour that is not seen in versions of RHEL 7 prior to RHEL 7.6
Environment
- Red Hat Enterprise Linux 7.6
- Red Hat Enterprise Linux as the NFS client
- kernel at least
3.10.0-957.el7
and before3.10.0-957.5.1.el7
- kernel at least
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.