[pnfs] Is Linux client <-> DS I/O working for anyone in the 2.6.26 tree?

[Marc Eshel]

Make sure your first request is not a big write the DS that will not work 
because of an NFS problem.
Marc.

pnfs-bounces at linux-nfs.org wrote on 06/13/2008 03:23:34 PM:

> "Muntz, Daniel" <Dan.Muntz at netapp.com> 
> Sent by: pnfs-bounces at linux-nfs.org
> 
> 06/13/2008 03:23 PM
> 
> To
> 
> "pNFS Mailing List" <pnfs at linux-nfs.org>
> 
> cc
> 
> Subject
> 
> [pnfs] Is Linux client <-> DS I/O working for anyone in the 2.6.26 tree?
> 
> Is Client <-> DS I/O working for anyone in the 2.6.26 tree?  Can anyone
> confirm that they've tried it successfully (or confirm failure for that
> matter...).  In our case, the DS is dropping the write request from the
> client:
> 
> Jun 13 15:07:55 quail kernel: nfsv4 compound op f6402850 opcnt 3 #2: 22:
> status 30000
> Jun 13 15:07:55 quail kernel: nfsd: Dropping request; may be revisited
> later
> Jun 13 15:07:55 quail kernel: svc: svc_process dropit
> Jun 13 15:07:55 quail kernel: svc: xprt f6d96400 dropped request
> Jun 13 15:07:55 quail kernel: svc: transport f6d96400 busy, not enqueued
> Jun 13 15:07:55 quail kernel: svc: server f6401000 waiting for data (to
> = 900000)
> Jun 13 15:07:55 quail kernel: svc: tcp_recv f6d96400 data 1 conn 0 close
> 0
> Jun 13 15:07:55 quail kernel: svc: socket f6d96400 recvfrom(f6d96580, 4)
> = -11
> Jun 13 15:07:55 quail kernel: RPC: TCP recvfrom got EAGAIN
> Jun 13 15:07:55 quail kernel: svc: got len=-11
> 
> Putting 2.6.26 on any of: client, DS, MDS, breaks things, but I don't
> have debugging info from the other scenarios yet--only the setup with
> all machines running 2.6.26.
> 
> Still debugging...
> 
>   -Dan
> _______________________________________________
> pNFS mailing list
> pNFS at linux-nfs.org
> http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs