Symptoms
In the NFSv3 file system, the df -h command is executed to query file system information.
Cause
Red Hat Enterprise Linux 7.6 kernel-3.10.0-957 and earlier kernel versions have bugs. After the TCP connection is disconnected, the first connection fails and the connection is repeatedly retried. The timeout duration is about 30s.
https://access.redhat.com/solutions/3742871
Analysis
1. Check whether the query result is displayed slowly every time or after a period of time (usually 5 minutes). If no, query other cases.
2. Run the netstat -ano command to check the countdown of the connection between the client and the logical port of the storage device. The value following keepalive is the countdown end time.
When the NFSv3 file system does not exchange packets for a period of time, the NFSv3 file system disconnects from the logical port of the storage device. Run the netstat –ano command to check whether the client is disconnected from the logical port of the storage device.
3. Ensure that each df -h freeze occurs after the connection is disconnected. In other scenarios, this problem is not caused by this problem. Refer to other cases to handle the problem.
4. According to the captured packets on the client, the TCP message fails to be sent. The connection is successful after five retries, which takes more than 30 seconds.
Solution
This problem is caused by an OS bug. You are advised to consult the OS vendor O&M engineers to determine the kernel version to which the problem is resolved.
https://access.redhat.com/solutions/3742871
