Hi, I have an NFS server running on Alpine Linux: # uname -a Linux charm 6.6.54-0-lts #1-Alpine SMP PREEMPT_DYNAMIC 2024-10-04 16:47:58 x86_64 Linux I wrote some very simple shell scripts to track & report NFS client usage. But I noticed that after some time, the TCP port reported in the "address" field of /proc/fs/nfsd/clients/*/info doesn't match the actual TCP port in use by the client. An example from the server side: # cat /proc/fs/nfsd/clients/4/info clientid: 0xdf43354e670522ab address: "192.168.93.218:752" status: confirmed seconds from last renew: 60 name: "Linux NFSv4.2 raspberrypi" minor version: 2 Implementation domain: "kernel.org" Implementation name: "Linux 6.1.21-v7+ #1642 SMP Mon Apr 3 17:20:52 BST 2023 armv7l" Implementation time: [0, 0] callback state: UP callback address: 192.168.93.218:0 # netstat -n | egrep '(Foreign|:2049)' Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 192.168.93.82:2049 192.168.93.218:776 ESTABLISHED tcp 0 0 192.168.93.82:2049 192.168.93.218:752 ESTABLISHED And on the client: % netstat -4n Active Internet connections (w/o servers) Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 192.168.93.218:776 192.168.93.82:2049 ESTABLISHED When the NFS connection was first established, it was using 752 (I had checked), but it seems that it got disconnected at some point and reconnected from port 776. (In this case I was actually doing some rewiring and the NFS server was temporarily disconnected from the network.) The client_info_show() function seems to be just reading the state data from a struct nfs4_client. And I see this gets populated in create_client() but I don't see it being updated anywhere. In this case, it would have to happen after the client reconnects following a transient error. Anyhow, I don't have an immediate solution, but wanted to mention this in case it wasn't already a known issue. Ideally the port could be updated when it changes so that it's possible to better track active clients. Many thanks, Tyler
