[ewg] NFSoRDMA: Can't restart NFS server

Jon Mason jon at opengridcomputing.com
Wed Dec 2 08:59:11 PST 2009


On Wed, Dec 02, 2009 at 07:44:30AM -0600, Jon Mason wrote:
> On Wed, Dec 02, 2009 at 06:00:03PM +0800, PN wrote:
> > Hi,
> > 
> > I'm using CentOS 5.4 with OFED-1.5-rc3 and Mellanox ConnectX cards.
> > I follow the steps of nfs-rdma.release-notes.txt and can use NFS over RDMA
> > successfully in the first time.
> > However, after umount the file system and restart the NFS server, it shows
> > the following error:
> > 
> > # /etc/init.d/nfs stop
> > Shutting down NFS mountd:                                  [  OK  ]
> > Shutting down NFS daemon:                                  [  OK  ]
> > Shutting down NFS quotas:                                  [  OK  ]
> > Shutting down NFS services:
> > 
> > # cat /proc/fs/nfsd/portlist
> > rdma 20049
> > tcp 2049
> > udp 2049
> > 
> > # lsmod |grep rdma
> > svcrdma                69184  1
> > sunrpc                245544  14 svcrdma,nfsd,nfs,nfs_acl,auth_rpcgss,lockd
> > rdma_ucm               47872  0
> > rdma_cm                69140  3 svcrdma,rdma_ucm,ib_sdp
> > iw_cm                  43656  1 rdma_cm
> > ib_addr                43016  1 rdma_cm
> > ib_cm                  73000  2 rdma_cm,ib_ipoib
> > ib_sa                  75144  3 rdma_cm,ib_ipoib,ib_cm
> > ib_uverbs              75696  1 rdma_ucm
> > ib_core               109312  15
> > svcrdma,rdma_ucm,ib_sdp,rdma_cm,iw_cm,ib_ipoib,ib_cm,ib_sa,ib_uverbs,ib_umad,iw_nes,iw_cxgb3,mlx4_ib,ib_mthca,ib_mad
> > 
> > # /etc/init.d/nfs start
> > Starting NFS services:                                     [  OK  ]
> > Starting NFS quotas:                                       [  OK  ]
> > Starting NFS daemon:                                       [FAILED]
> 
> You want to run the following to restart:
> # /etc/init.d/nfs restart
> 
> > In /var/log/messages, it shows:
> > Dec  2 17:34:30 mgmt nfsd[5106]: nfssvc: Setting version failed: errno 16
> > (Device or resource busy)
> > Dec  2 17:34:30 mgmt nfsd[5106]: nfssvc: unable to bind UPD socket: errno 98
> > (Address already in use)
> > 
> > And in the client side, I can no longer mount the server again.
> 
> This could be a bug, I'll look into it.

I can replicate the issue running /etc/init.d/nfs restart on a CentOS
5.3 server and a CentOS 5.4 client.  Please open a bug.

Thanks,
Jon

> 
> > Is there a correct ways to restart the NFS server after configuring
> > NFSoRDMA?
> > 
> > Thanks in advance.
> > 
> > PN
> 
> > _______________________________________________
> > ewg mailing list
> > ewg at lists.openfabrics.org
> > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
> _______________________________________________
> ewg mailing list
> ewg at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg



More information about the ewg mailing list