We are experiencing a NFS v3 lockd duplicate SYN that looks related to https://access.redhat.com/solutions/3765711 and/or https://access.redhat.com/solutions/3742871 which states: The issue (bz1660784) has been resolved with errata RHSA-2019:0163 with the following package(s): kernel-3.10.0-957.5.1.el7 or later.
We updated to kernel 3.10.0-957.10.1 from 3.10.0-1.957.3.el7 and that is when we starting having this intermittent issue.
Only have seen this on RHEL 7.6 client to ISILON NFS servers after we updated from Kernel 3.10.0-957.1.3 to 3.10.0-957.10.1. Before this update, the client might send an RST in the middle of connection setup but we could recover from that.
We can not reproduce it at will / it does not always send a duplicate SYN .. and now we keep the session alive
I did notice in one of my older traces that if the client tore down the connection after 5 minute of idle time then the next lockd request would create the session ; but if the Isilon tore down the connection then that is when the client would send the duplicate SYN packets and we would never recover after that.
- Red Hat Enterprise Linux 7.6 (NFS client)
- kernels 3.10.0-957.5.1.el7 or above
- Red Hat Enterprise Linux 8.0 (NFS client)
- kernel 4.18.0-80.el8
- seen with Isilon OneFS v18.104.22.168 B_8_1_2_016 (NFS server)
- NOTE: only believed to be due to the shorter idle timer, not a bug on this server
- stateful firewalls and firepower IDS/IPS between NFS client and server
- NFS server initiated shutdown of TCP connection (ESTABLISHED, CLOSE-WAIT, LAST-ACK, CLOSED)
- seen with NFSv3
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase of over 48,000 articles and solutions.