> On Jun 20, 2024, at 1:29 AM, Harald Dunkel <harald.dunkel@xxxxxxxxxx> wrote: > > Hi Chuck, > > On 2024-06-19 15:14:15, Chuck Lever III wrote: >> There's no way to answer either of these questions >> since we have no idea what the problem is yet. > > I am getting these messages again, similar to previous > incidents: > > [Wed Jun 19 16:46:14 2024] receive_cb_reply: Got unrecognized reply: calldir 0x1 xpt_bc_xprt 00000000a5ff94a6 xid f9d6a568 > [Wed Jun 19 18:42:23 2024] receive_cb_reply: Got unrecognized reply: calldir 0x1 xpt_bc_xprt 00000000a5ff94a6 xid 59d8a568 > [Wed Jun 19 18:43:15 2024] receive_cb_reply: Got unrecognized reply: calldir 0x1 xpt_bc_xprt 00000000a5ff94a6 xid 5cd8a568 > > nfsd is still running as usual. Assuming they are related to > NFS, Is there a way to map these weird messages to a remote IP > address? I could setup a similar system and use it to capture > the traffic without breaking privacy. These messages do come from kernel NFS code, but it's still not possible to say how they are related to the hang/deadlock. The best you can do here is enable server tracing: $ sudo su - # trace-cmd record -e nfsd:nfsd_cb\* That will pick up NFSv4 callback related events. When you are done tracing, ^C. # trace-cmd report | less For trace points that record client address information, you might also need the -R option to see everything in each trace record. -- Chuck Lever
