tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1################################################################################
2# #
3# NFS/RDMA README #
4# #
5################################################################################
6
7 Author: NetApp and Open Grid Computing
8 Date: April 15, 2008
9
10Table of Contents
11~~~~~~~~~~~~~~~~~
12 - Overview
13 - Getting Help
14 - Installation
15 - Check RDMA and NFS Setup
16 - NFS/RDMA Setup
17
18Overview
19~~~~~~~~
20
21 This document describes how to install and setup the Linux NFS/RDMA client
22 and server software.
23
24 The NFS/RDMA client was first included in Linux 2.6.24. The NFS/RDMA server
25 was first included in the following release, Linux 2.6.25.
26
27 In our testing, we have obtained excellent performance results (full 10Gbit
28 wire bandwidth at minimal client CPU) under many workloads. The code passes
29 the full Connectathon test suite and operates over both Infiniband and iWARP
30 RDMA adapters.
31
32Getting Help
33~~~~~~~~~~~~
34
35 If you get stuck, you can ask questions on the
36
37 nfs-rdma-devel@lists.sourceforge.net
38
39 mailing list.
40
41Installation
42~~~~~~~~~~~~
43
44 These instructions are a step by step guide to building a machine for
45 use with NFS/RDMA.
46
47 - Install an RDMA device
48
49 Any device supported by the drivers in drivers/infiniband/hw is acceptable.
50
51 Testing has been performed using several Mellanox-based IB cards, the
52 Ammasso AMS1100 iWARP adapter, and the Chelsio cxgb3 iWARP adapter.
53
54 - Install a Linux distribution and tools
55
56 The first kernel release to contain both the NFS/RDMA client and server was
57 Linux 2.6.25 Therefore, a distribution compatible with this and subsequent
58 Linux kernel release should be installed.
59
60 The procedures described in this document have been tested with
61 distributions from Red Hat's Fedora Project (http://fedora.redhat.com/).
62
63 - Install nfs-utils-1.1.1 or greater on the client
64
65 An NFS/RDMA mount point can only be obtained by using the mount.nfs
66 command in nfs-utils-1.1.1 or greater. To see which version of mount.nfs
67 you are using, type:
68
69 > /sbin/mount.nfs -V
70
71 If the version is less than 1.1.1 or the command does not exist,
72 then you will need to install the latest version of nfs-utils.
73
74 Download the latest package from:
75
76 http://www.kernel.org/pub/linux/utils/nfs
77
78 Uncompress the package and follow the installation instructions.
79
80 If you will not be using GSS and NFSv4, the installation process
81 can be simplified by disabling these features when running configure:
82
83 > ./configure --disable-gss --disable-nfsv4
84
85 For more information on this see the package's README and INSTALL files.
86
87 After building the nfs-utils package, there will be a mount.nfs binary in
88 the utils/mount directory. This binary can be used to initiate NFS v2, v3,
89 or v4 mounts. To initiate a v4 mount, the binary must be called mount.nfs4.
90 The standard technique is to create a symlink called mount.nfs4 to mount.nfs.
91
92 NOTE: mount.nfs and therefore nfs-utils-1.1.1 or greater is only needed
93 on the NFS client machine. You do not need this specific version of
94 nfs-utils on the server. Furthermore, only the mount.nfs command from
95 nfs-utils-1.1.1 is needed on the client.
96
97 - Install a Linux kernel with NFS/RDMA
98
99 The NFS/RDMA client and server are both included in the mainline Linux
100 kernel version 2.6.25 and later. This and other versions of the 2.6 Linux
101 kernel can be found at:
102
103 ftp://ftp.kernel.org/pub/linux/kernel/v2.6/
104
105 Download the sources and place them in an appropriate location.
106
107 - Configure the RDMA stack
108
109 Make sure your kernel configuration has RDMA support enabled. Under
110 Device Drivers -> InfiniBand support, update the kernel configuration
111 to enable InfiniBand support [NOTE: the option name is misleading. Enabling
112 InfiniBand support is required for all RDMA devices (IB, iWARP, etc.)].
113
114 Enable the appropriate IB HCA support (mlx4, mthca, ehca, ipath, etc.) or
115 iWARP adapter support (amso, cxgb3, etc.).
116
117 If you are using InfiniBand, be sure to enable IP-over-InfiniBand support.
118
119 - Configure the NFS client and server
120
121 Your kernel configuration must also have NFS file system support and/or
122 NFS server support enabled. These and other NFS related configuration
123 options can be found under File Systems -> Network File Systems.
124
125 - Build, install, reboot
126
127 The NFS/RDMA code will be enabled automatically if NFS and RDMA
128 are turned on. The NFS/RDMA client and server are configured via the hidden
129 SUNRPC_XPRT_RDMA config option that depends on SUNRPC and INFINIBAND. The
130 value of SUNRPC_XPRT_RDMA will be:
131
132 - N if either SUNRPC or INFINIBAND are N, in this case the NFS/RDMA client
133 and server will not be built
134 - M if both SUNRPC and INFINIBAND are on (M or Y) and at least one is M,
135 in this case the NFS/RDMA client and server will be built as modules
136 - Y if both SUNRPC and INFINIBAND are Y, in this case the NFS/RDMA client
137 and server will be built into the kernel
138
139 Therefore, if you have followed the steps above and turned no NFS and RDMA,
140 the NFS/RDMA client and server will be built.
141
142 Build a new kernel, install it, boot it.
143
144Check RDMA and NFS Setup
145~~~~~~~~~~~~~~~~~~~~~~~~
146
147 Before configuring the NFS/RDMA software, it is a good idea to test
148 your new kernel to ensure that the kernel is working correctly.
149 In particular, it is a good idea to verify that the RDMA stack
150 is functioning as expected and standard NFS over TCP/IP and/or UDP/IP
151 is working properly.
152
153 - Check RDMA Setup
154
155 If you built the RDMA components as modules, load them at
156 this time. For example, if you are using a Mellanox Tavor/Sinai/Arbel
157 card:
158
159 > modprobe ib_mthca
160 > modprobe ib_ipoib
161
162 If you are using InfiniBand, make sure there is a Subnet Manager (SM)
163 running on the network. If your IB switch has an embedded SM, you can
164 use it. Otherwise, you will need to run an SM, such as OpenSM, on one
165 of your end nodes.
166
167 If an SM is running on your network, you should see the following:
168
169 > cat /sys/class/infiniband/driverX/ports/1/state
170 4: ACTIVE
171
172 where driverX is mthca0, ipath5, ehca3, etc.
173
174 To further test the InfiniBand software stack, use IPoIB (this
175 assumes you have two IB hosts named host1 and host2):
176
177 host1> ifconfig ib0 a.b.c.x
178 host2> ifconfig ib0 a.b.c.y
179 host1> ping a.b.c.y
180 host2> ping a.b.c.x
181
182 For other device types, follow the appropriate procedures.
183
184 - Check NFS Setup
185
186 For the NFS components enabled above (client and/or server),
187 test their functionality over standard Ethernet using TCP/IP or UDP/IP.
188
189NFS/RDMA Setup
190~~~~~~~~~~~~~~
191
192 We recommend that you use two machines, one to act as the client and
193 one to act as the server.
194
195 One time configuration:
196
197 - On the server system, configure the /etc/exports file and
198 start the NFS/RDMA server.
199
200 Exports entries with the following formats have been tested:
201
202 /vol0 192.168.0.47(fsid=0,rw,async,insecure,no_root_squash)
203 /vol0 192.168.0.0/255.255.255.0(fsid=0,rw,async,insecure,no_root_squash)
204
205 The IP address(es) is(are) the client's IPoIB address for an InfiniBand HCA or the
206 cleint's iWARP address(es) for an RNIC.
207
208 NOTE: The "insecure" option must be used because the NFS/RDMA client does not
209 use a reserved port.
210
211 Each time a machine boots:
212
213 - Load and configure the RDMA drivers
214
215 For InfiniBand using a Mellanox adapter:
216
217 > modprobe ib_mthca
218 > modprobe ib_ipoib
219 > ifconfig ib0 a.b.c.d
220
221 NOTE: use unique addresses for the client and server
222
223 - Start the NFS server
224
225 If the NFS/RDMA server was built as a module (CONFIG_SUNRPC_XPRT_RDMA=m in kernel config),
226 load the RDMA transport module:
227
228 > modprobe svcrdma
229
230 Regardless of how the server was built (module or built-in), start the server:
231
232 > /etc/init.d/nfs start
233
234 or
235
236 > service nfs start
237
238 Instruct the server to listen on the RDMA transport:
239
240 > echo rdma 2050 > /proc/fs/nfsd/portlist
241
242 - On the client system
243
244 If the NFS/RDMA client was built as a module (CONFIG_SUNRPC_XPRT_RDMA=m in kernel config),
245 load the RDMA client module:
246
247 > modprobe xprtrdma.ko
248
249 Regardless of how the client was built (module or built-in), issue the mount.nfs command:
250
251 > /path/to/your/mount.nfs <IPoIB-server-name-or-address>:/<export> /mnt -i -o rdma,port=2050
252
253 To verify that the mount is using RDMA, run "cat /proc/mounts" and check the
254 "proto" field for the given mount.
255
256 Congratulations! You're using NFS/RDMA!