Re: nfs stalls over loopback interface (no sk_data_ready events?)

[Jeff Layton <jeff.layton-7I+n7zu2hftEKMMhf/gKZA@public.gmane.org>]

On Mon, 27 Oct 2014 15:29:00 -0400
Jeff Layton <jlayton-7I+n7zu2hftEKMMhf/gKZA@public.gmane.org> wrote:

> (sorry for resend -- I got the netdev address wrong)
> 
> Sending this to netdev since I think I've now determined that this is
> not a NFS specific problem. Recently Christoph mentioned that he was
> seeing stalls when running xfstests generic/075 test on NFS over the
> loopback interface with v3.18-rc1-ish kernel.
> 
> The configuration in this case is the nfs server and client on same box
> communicating over the lo interface.
> 
> Here's are tracepoints from a typical request as it's supposed to work:
> 
>        mount.nfs-906   [002] ...1 22711.996969: xprt_transmit: xprt=0xffff8800ce961000 xid=0xa8a34513 status=0
>             nfsd-678   [000] ...1 22711.997082: svc_recv: rq_xid=0xa8a34513 status=164
>             nfsd-678   [000] ..s8 22711.997185: xprt_lookup_rqst: xprt=0xffff8800ce961000 xid=0xa8a34513 status=0
>             nfsd-678   [000] ..s8 22711.997186: xprt_complete_rqst: xprt=0xffff8800ce961000 xid=0xa8a34513 status=140
>             nfsd-678   [000] ...1 22711.997236: svc_send: rq_xid=0xa8a34513 dropme=0 status=144
>             nfsd-678   [000] ...1 22711.997236: svc_process: rq_xid=0xa8a34513 dropme=0 status=144
> 
> ...basically, we send a request to the server. Server picks it up and
> sends the reply, and then the client IDs that reply and processes it.
> This runs along just fine for ~ a minute or so. At some point, the
> client stops seeing replies come in:
> 
>      kworker/2:2-107   [002] ...1 22741.696070: xprt_transmit: xprt=0xffff8800ce961000 xid=0xc3a84513 status=0
>             nfsd-678   [002] .N.1 22741.696917: svc_recv: rq_xid=0xc3a84513 status=208
>             nfsd-678   [002] ...1 22741.699890: svc_send: rq_xid=0xc3a84513 dropme=0 status=262252
>             nfsd-678   [002] ...1 22741.699891: svc_process: rq_xid=0xc3a84513 dropme=0 status=262252
> 
> 
> ...a bit more tracepoint work seems to show that we just stop getting
> sk_data_ready callbacks on the socket at all. I'm not terribly familiar
> with the lower-level socket code, so I figured I'd email here and ask...
> 
> Anyone have insight into why this might be happening?
> 

Looks some change that went into -rc2 has fixed the problem for me.
Christoph, can you confirm that this no longer occurs with -rc2?

-- 
Jeff Layton <jlayton-7I+n7zu2hftEKMMhf/gKZA@public.gmane.org>
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html