RHEL 7 NFS issues (xs_tcp_setup_socket: connect returned unhandled error -107)
Greetings,
I am having some issues with NFS clients on two separate physical servers. Often 'dmesg -T' shows the following output several times an hour:
dmesg:
xs_tcp_setup_socket: connect returned unhandled error -107
/var/log/messages:
kernel: xs_tcp_setup_socket: connect returned unhandled error -107
This is apparently causing applications to hang for a time. I am unable to find any information regarding this error specifically for RHEL7. The server is a NetApp appliance with dozens of other NFS shares. We've tried re-creating the shares, just in case... and also connecting to shares we know are working, with the same results.
Anyone else having this issue? We're pinned to RHEL 7.3.
A further note: This is a production server; the dev and test servers running the same application, same OS aren't having this issue.
Distributor ID: RedHatEnterpriseServer
Release: 7.3
Linux hostname 3.10.0-514.26.2.el7.x86_64 #1 SMP Fri Jun 30 05:26:04 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
Responses
Red Hat
Guru
7257 points
Error number 107 is ENOTCONN 107 /* Transport endpoint is not connected */
Following the logic in the RPC client, xs_tcp_setup_socket() calls xs_tcp_finish_connecting() and ENOTCONN is the default return.
The logic in the RPC client doesn't make sense to me here - creating a transport with port 0 is legal (asking the kernel to assign a port) but then if the port number is 0 at connect time, skip connecting and return this error? The only thing I can think is that you have so many ports open to this NFS Server that the client can't make new ports, but that seems unlikely. The RPC client isn't an area I cover so I'm definitely not the best person to ask, there's obviously something I've missed.
You have technical support entitlements so this would be a good support case.
Put "nfs" in the case subject so it is routed to the correct team. Follow the "Protocol Issues" troubleshooting path of How to begin NFS Debugging which will enable RPC debugging, then reproduce the error (dmesg will have a LOT more data now). Include a sosreport from an affected client, which will pull in the messages file with the new RPC debugging around the -107 error message you describe.
That will get the case off to a good start.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.