| 1 | {{{ |
| 2 | =============================== |
| 3 | Tutorial for Tahoe-LAFS (Draft) |
| 4 | =============================== |
| 5 | }}} |
| 6 | = Hosts setup = |
| 7 | |
| 8 | In this tutorial we're gonna setup 3 nodes, whatever they are (Host+VM1+VM2) |
| 9 | |
| 10 | IPs and hosts are: |
| 11 | * 172.16.23.1 (host.local : Mac OS X 10.6.4) |
| 12 | * 172.16.23.128 (vm1.local : Debian GNU/Linux 5.0) |
| 13 | * 172.16.23.130 (vm2.local : Debian GNU/Linux 5.0) |
| 14 | |
| 15 | I'll use a self-explanatory prompt for each host. |
| 16 | |
| 17 | = Installation on each node = |
| 18 | |
| 19 | I like pypi so much I find it easier to do the download, extraction, |
| 20 | get dependencies and build using it. |
| 21 | {{{ |
| 22 | # easy_install -U allmydata-tahoe |
| 23 | }}} |
| 24 | it failed on the freshly installed debian, with that error : |
| 25 | {{{ |
| 26 | ImportError: No module named darcsver.setuptools_command |
| 27 | }}} |
| 28 | A fast look on google, showed that installing setuptools_darcs and |
| 29 | darcsver don't build as expected and may be missing at install time. |
| 30 | The fix is to install them first, and only then install tahoe. |
| 31 | {{{ |
| 32 | # easy_install -U setuptools_darcs darcsver |
| 33 | # easy_install -U allmydata-tahoe |
| 34 | }}} |
| 35 | then execute tahoe --vesion to check everything is installed as |
| 36 | expected : |
| 37 | |
| 38 | on debian : |
| 39 | {{{ |
| 40 | % tahoe --version |
| 41 | allmydata-tahoe: 1.7.1, foolscap: 0.5.1, pycryptopp: 0.5.17-r683, zfec: 1.4.7, Twisted: 10.0.0, Nevow: 0.10.0, zope.interface: 3.5.1, python: 2.6.1, platform: Darwin-10.4.0-i386-64bit, sqlite: 3.6.12, simplejson: 2.1.1, argparse: 1.1, pycrypto: 2.1.0, pyOpenSSL: 0.7, pyutil: 1.7.7, zbase32: 1.1.1, setuptools: 0.6c15dev, pyasn1: 0.0.11a, pysqlite: 2.4.1 |
| 42 | }}} |
| 43 | on MacOSX : |
| 44 | {{{ |
| 45 | % tahoe --version |
| 46 | allmydata-tahoe: 1.7.1, foolscap: 0.5.1, pycryptopp: 0.5.19, zfec: 1.4.7, Twisted: 10.0.0, Nevow: 0.10.0, zope.interface: 3.6.1, python: 2.5.2, platform: Linux-debian_5.0-x86_64-64bit, sqlite: 3.5.9, simplejson: 2.1.1, argparse: 1.1, pycrypto: 2.1.0, pyOpenSSL: 0.7, pyutil: 1.7.7, zbase32: 1.1.1, setuptools: 0.6c12dev, pyasn1: 0.0.11a, pysqlite: 2.3.2 |
| 47 | }}} |
| 48 | now Tahoe is installed \o/ |
| 49 | |
| 50 | = Instanciate the nodes = |
| 51 | |
| 52 | So, after reading running.html, configuration.txt and architecture.txt, we'll first |
| 53 | create an introducer node, that will get every other nodes to know each other. The |
| 54 | documentation says that we have to create and launch the introducer as follows (that |
| 55 | we do on vm1.local) |
| 56 | {{{ |
| 57 | vm1% mkdir tahoe-introducer |
| 58 | vm1% tahoe create-introducer tahoe-introducer |
| 59 | Introducer created in tahoe-introducer |
| 60 | vm1% tahoe start tahoe-introducer |
| 61 | STARTING /home/guyzmo/tahoe-introducer |
| 62 | introducer node probably started |
| 63 | vm1% cat tahoe-introducer/introducer.furl |
| 64 | pb://6oibvp5whrb3v3ge7ajuyapooswsda2e@172.16.23.128:33272,127.0.0.1:33272/introducer |
| 65 | }}} |
| 66 | Then we copy that last string we got from 'cat', and paste it in tahoe.cfg, |
| 67 | as follows (edit it using your favorite editor, mine is vim) : |
| 68 | {{{ |
| 69 | vm2% vim ~/.tahoe/tahoe.cfg |
| 70 | -------------8<----------------------8<---------------- |
| 71 | 22 [client] |
| 72 | 23 introducer.furl = pb://6oibvp5whrb3v3ge7ajuyapooswsda2e@172.16.23.128:33272,127.0.0.1:33272/introducer |
| 73 | 24 helper.furl = |
| 74 | 25 #key_generator.furl = |
| 75 | 26 #stats_gatherer.furl = |
| 76 | 27 #shares.needed = 3 |
| 77 | 28 #shares.happy = 7 |
| 78 | 29 #shares.total = 10 |
| 79 | ------------->8---------------------->8---------------- |
| 80 | }}} |
| 81 | exit, and then do the same on 'host'. |
| 82 | |
| 83 | finally, execute on 'vm2' and 'host' the following : |
| 84 | {{{ |
| 85 | vm2% tahoe start |
| 86 | STARTING /home/guyzmo/.tahoe |
| 87 | client node probably started |
| 88 | |
| 89 | host% tahoe start |
| 90 | STARTING /home/guyzmo/.tahoe |
| 91 | client node probably started |
| 92 | }}} |
| 93 | wait a bit and do : |
| 94 | {{{ |
| 95 | % ps aux | grep tahoe |
| 96 | myuser 5360 1.2 7.4 158820 28348 ? S< 02:13 0:00 /usr/bin/python /usr/bin/twistd -y tahoe-client.tac --logfile logs/twistd.log |
| 97 | }}} |
| 98 | and you shall see an output that looks alike. If not, tahoe failed to start. |
| 99 | You've got to check ~/.tahoe/logs/ what got wrong. |
| 100 | |
| 101 | = Configure the nodes = |
| 102 | |
| 103 | Now, let's do some configuration. On each node edit tahoe.cfg : |
| 104 | {{{ |
| 105 | host% vim ~/.tahoe/tahoe.cfg |
| 106 | -------------8<----------------------8<---------------- |
| 107 | 10 [node] |
| 108 | 11 nickname = client-host |
| 109 | 12 web.port = tcp:3456:interface=172.16.23.1 |
| 110 | ------------->8---------------------->8---------------- |
| 111 | host% tahoe restart |
| 112 | }}} |
| 113 | |
| 114 | {{{ |
| 115 | vm1% vim ~/tahoe-introducer/tahoe.cfg |
| 116 | -------------8<----------------------8<---------------- |
| 117 | 10 [node] |
| 118 | 11 nickname = introducer-vm1 |
| 119 | 12 web.port = tcp:3456:interface=172.16.23.128 |
| 120 | ------------->8---------------------->8---------------- |
| 121 | vm1% tahoe restart ~/tahoe-introducer |
| 122 | |
| 123 | vm2% vim ~/.tahoe/tahoe.cfg |
| 124 | -------------8<----------------------8<---------------- |
| 125 | 10 [node] |
| 126 | 11 nickname = client-vm2 |
| 127 | 12 web.port = tcp:3456:interface=172.16.23.130 |
| 128 | ------------->8---------------------->8---------------- |
| 129 | vm2% tahoe restart |
| 130 | }}} |
| 131 | |
| 132 | Then get your favorite browser and browse to the following URLs, you |
| 133 | should get pages similar to the ones below |
| 134 | |
| 135 | * http://172.16.23.1:3456/ -> http://m0g.net/~guyzmo/172.16.23.1_3456_1.png |
| 136 | * http://172.16.23.128:3456/ -> http://m0g.net/~guyzmo/172.16.23.128_3456_1.png |
| 137 | * http://172.16.23.130:3456/ -> http://m0g.net/~guyzmo/172.16.23.130_3456_1.png |
| 138 | |
| 139 | Now it's time to be able to store something on the grid, as |
| 140 | the whole purpose of Tahoe-LAFS, and thus of this tutorial is |
| 141 | to get a grid to store files. |
| 142 | |
| 143 | On host and vm2, edit again ~/.tahoe/tahoe.cfg and modify it |
| 144 | as follows : |
| 145 | |
| 146 | {{{ |
| 147 | vm2% vim ~/.tahoe/tahoe.cfg |
| 148 | -------------8<----------------------8<---------------- |
| 149 | 31 [storage] |
| 150 | 32 enabled = true |
| 151 | 33 readonly = false |
| 152 | 34 reserved_space = 1GB # here you tell the storage server how much space you would like to use |
| 153 | ------------->8---------------------->8---------------- |
| 154 | vm2% tahoe restart |
| 155 | }}} |
| 156 | |
| 157 | and the same on host. |
| 158 | |
| 159 | The result would be as follows : |
| 160 | |
| 161 | * http://172.16.23.1:3456/ -> http://m0g.net/~guyzmo/172.16.23.1_3456_2.png |
| 162 | * http://172.16.23.128:3456/ -> http://m0g.net/~guyzmo/172.16.23.128_3456_2.png |
| 163 | * http://172.16.23.130:3456/ -> http://m0g.net/~guyzmo/172.16.23.130_3456_2.png |
| 164 | |
| 165 | Then, you've got a working Tahoe-LAFS grid working ! |
| 166 | Is that all ? Well, no, it's time to learn how to use it. |
| 167 | |
| 168 | = File capabilities ? = |
| 169 | |
| 170 | Now it's time for a bit of litterature. |
| 171 | |
| 172 | TODO |
| 173 | |
| 174 | = Access your shares = |
| 175 | |
| 176 | You can store and access files on Tahoe-LAFS using a lot of |
| 177 | different ways. The fatest is to either use the Web UI, or |
| 178 | the CLI though you need to remember the caps of the files. |
| 179 | |
| 180 | Using the WUI is straight forward, connect to a storage client, |
| 181 | use the forms to add/get/remove files and directories. Just |
| 182 | always remember that once you created a directory, you need to |
| 183 | save the file's URI, or you won't be able to find it again later |
| 184 | on. |
| 185 | |
| 186 | == Using the CLI == |
| 187 | |
| 188 | The CLI behaves mostly like a standard shell, with all |
| 189 | commands prefixed by 'tahoe', the same way version control |
| 190 | system work. |
| 191 | |
| 192 | If you execute for the first time : |
| 193 | |
| 194 | {{{ |
| 195 | % tahoe ls |
| 196 | error: No alias specified, and the default 'tahoe' alias doesn't exist. To create it, use 'tahoe create-alias tahoe'. |
| 197 | }}} |
| 198 | |
| 199 | you'll get that error. That's why you need to create a |
| 200 | 'root' directory for CLI commands that will be aliased |
| 201 | to the 'tahoe:' prefix. |
| 202 | |
| 203 | {{{ |
| 204 | % tahoe create-alias tahoe |
| 205 | [...] |
| 206 | socket.error: [Errno 61] Connection refused |
| 207 | }}} |
| 208 | |
| 209 | What happens here is that the Tahoe client tries to connect to the URL |
| 210 | given in the ~/.tahoe/node.url file. So, you need to edit ~/.tahoe/node.url |
| 211 | {{{ |
| 212 | host% vim ~/.tahoe/node.url |
| 213 | -------------8<----------------------8<---------------- |
| 214 | 1 http://172.16.23.1:3456/ |
| 215 | ------------->8---------------------->8---------------- |
| 216 | }}} |
| 217 | {{{ |
| 218 | vm2% vim ~/.tahoe/node.url |
| 219 | -------------8<----------------------8<---------------- |
| 220 | 1 http://172.16.23.1:3456/ |
| 221 | ------------->8---------------------->8---------------- |
| 222 | }}} |
| 223 | |
| 224 | HINT: each time you do 'tahoe start' node.url gets reset to 127.0.0.1. If you |
| 225 | want to intensively use the CLI, change |
| 226 | 'web.port = tcp:3456:interface=172.16.23.1' |
| 227 | to |
| 228 | 'web.port = tcp:3456' |
| 229 | in tahoe.cfg. Specifying the interface to listen on is usually a good security |
| 230 | policy, though twisted supports only one interface declaration, or all interfaces |
| 231 | if you don't declare it. |
| 232 | |
| 233 | Now, you can run the previous command : |
| 234 | |
| 235 | {{{ |
| 236 | % tahoe create-alias tahoe |
| 237 | Alias 'tahoe' created |
| 238 | }}} |
| 239 | |
| 240 | What it has actually done, is that it edited the ~/.tahoe/private/aliases file |
| 241 | linking the keyword you gave with a file cap : |
| 242 | |
| 243 | {{{ |
| 244 | vm2% cat ~/.tahoe/private/aliases |
| 245 | tahoe: URI:DIR2:bfe7inwcduobfnnrxocqechr4q:huurbpkrtajzdrq6okvxxppj5negccz4payh6gxes5nphe6djbeq |
| 246 | }}} |
| 247 | and |
| 248 | {{{ |
| 249 | host% cat ~/.tahoe/private/aliases |
| 250 | tahoe: URI:DIR2:ozgebrm5oe52hhcnxhbc6oy6xu:7cyu73izd6q6w4kwngyxxdu7ttuzwy5kl2uybcrqohhen362crla |
| 251 | }}} |
| 252 | so, as you can see, different tahoe CLI won't work in the same base directories. |
| 253 | If you really want to enable this, copy the same aliases file on every account |
| 254 | you want to use the CLI on. |
| 255 | |
| 256 | Finally, to use the CLI, have a look at 'tahoe help' for a full description |
| 257 | of the commands. |
| 258 | |
| 259 | For example, here are some basic commands: |
| 260 | |
| 261 | {{{ |
| 262 | % tahoe cp README.txt tahoe: |
| 263 | Success: file copied |
| 264 | % tahoe ls -l tahoe: |
| 265 | -r-- 1348 Aug 04 12:52 README.txt |
| 266 | % tahoe get tahoe:README.txt | head |
| 267 | =============================== |
| 268 | Tahoe-LAFS |
| 269 | =============================== |
| 270 | |
| 271 | Tahoe-LAFS is a Free Software/Open Source decentralized data store. It |
| 272 | distributes your filesystem across multiple servers, and even if some of the |
| 273 | servers fail or are taken over by an attacker, the entire filesystem continues |
| 274 | to work correctly and to preserve your privacy and security. |
| 275 | |
| 276 | To get started please see `quickstart.html`_. |
| 277 | }}} |
| 278 | |
| 279 | == Using the FTP interface == |
| 280 | |
| 281 | As Tahoe does not have user accounts management, to enable FTP access, |
| 282 | we need first to enable authentication and associate root caps to each |
| 283 | user allowed to access the system. |
| 284 | |
| 285 | /!\ Be sure to have installed twisted version >=10.1 before trying FTP/sFTP |
| 286 | interface, tahoe's ftp feature is not compatible with prior versions. |
| 287 | |
| 288 | For more informations about FTP/sFTP configuration refer to the documentation: |
| 289 | * http://tahoe-lafs.org/source/tahoe-lafs/trunk/docs/frontends/FTP-and-SFTP.txt |
| 290 | * http://tahoe-lafs.org/trac/tahoe-lafs/wiki/SftpFrontend |
| 291 | |
| 292 | === Account file creation === |
| 293 | |
| 294 | The password file is using a very simple format, each non blank/commented |
| 295 | line is following this template : |
| 296 | {{{ |
| 297 | USER PASSWORD ROOTCAP |
| 298 | }}} |
| 299 | so we're gonna setup two users the following way on host: |
| 300 | {{{ |
| 301 | host% tahoe create-alias alice |
| 302 | Alias 'alice' created |
| 303 | host% tahoe create-alias bob |
| 304 | Alias 'bob' created |
| 305 | host% cat ~/.tahoe/private/aliases |
| 306 | tahoe: URI:DIR2:ozgebrm5oe52hhcnxhbc6oy6xu:7cyu73izd6q6w4kwngyxxdu7ttuzwy5kl2uybcrqohhen362crla |
| 307 | alice: URI:DIR2:dhl2wcuoppuirtv3gk2lslhx6i:nur5sar5lwfxal63gcjaysu4mxt33l35uw4bqtokby3airqhwehq |
| 308 | bob: URI:DIR2:r2dwmysrl4wigiov52fmifs2mi:2fdu2l3jxuqupepefmkd3y5vftqatc62vl6elmb5qlgi3gmvpx6a |
| 309 | host% cat > ~/.tahoe/private/ftp.accounts |
| 310 | alice h3rp4ssw0rd URI:DIR2:dhl2wcuoppuirtv3gk2lslhx6i:nur5sar5lwfxal63gcjaysu4mxt33l35uw4bqtokby3airqhwehq |
| 311 | bob h1sp4ssw0rd URI:DIR2:r2dwmysrl4wigiov52fmifs2mi:2fdu2l3jxuqupepefmkd3y5vftqatc62vl6elmb5qlgi3gmvpx6a |
| 312 | host% |
| 313 | }}} |
| 314 | and two users on vm2: |
| 315 | {{{ |
| 316 | vm2% tahoe create-alias alice |
| 317 | Alias 'alice' created |
| 318 | vm2% tahoe add-alias bob URI:DIR2:r2dwmysrl4wigiov52fmifs2mi:2fdu2l3jxuqupepefmkd3y5vftqatc62vl6elmb5qlgi3gmvpx6a |
| 319 | vm2% cat ~/.tahoe/private/aliases |
| 320 | tahoe: URI:DIR2:ozgebrm5oe52hhcnxhbc6oy6xu:7cyu73izd6q6w4kwngyxxdu7ttuzwy5kl2uybcrqohhen362crla |
| 321 | alice: URI:DIR2:x3y3i3s6h5l7gspse6ol4b7goy:zinwdjlffv23gbv5umyjqcdg3trfmvxdwc77cwae3cpxw6mhtq5a |
| 322 | bob: URI:DIR2:r2dwmysrl4wigiov52fmifs2mi:2fdu2l3jxuqupepefmkd3y5vftqatc62vl6elmb5qlgi3gmvpx6a |
| 323 | vm2% cat > ~/.tahoe/private/ftp.accounts |
| 324 | alice h3rp4ssw0rd URI:DIR2:x3y3i3s6h5l7gspse6ol4b7goy:zinwdjlffv23gbv5umyjqcdg3trfmvxdwc77cwae3cpxw6mhtq5a |
| 325 | bob h1sp4ssw0rd URI:DIR2:r2dwmysrl4wigiov52fmifs2mi:2fdu2l3jxuqupepefmkd3y5vftqatc62vl6elmb5qlgi3gmvpx6a |
| 326 | vm2% |
| 327 | }}} |
| 328 | |
| 329 | If you look carefully to both logs, alice gets two root caps on each target host, |
| 330 | while bob is getting the same root cap on both. Of course, the alias creation is |
| 331 | optional, but handy for the grid manager to control what the users have. |
| 332 | |
| 333 | now, let's update the configuration : |
| 334 | {{{ |
| 335 | % vim ~/.tahoe/tahoe.cfg |
| 336 | -------------8<----------------------8<---------------- |
| 337 | 22 [ftpd] |
| 338 | 23 enabled = true |
| 339 | 24 port = 8021 |
| 340 | 25 accounts.file = private/ftp.accounts |
| 341 | ------------->8---------------------->8---------------- |
| 342 | % tahoe restart |
| 343 | STOPPING /home/guyzmo/.tahoe |
| 344 | process 3063 is dead |
| 345 | STARTING /home/guyzmo/.tahoe |
| 346 | client node probably started |
| 347 | }}} |
| 348 | |
| 349 | and then, connect your FTP client to tahoe's FTP server: |
| 350 | |
| 351 | {{{ |
| 352 | % tahoe cp foobar.jpg bob: |
| 353 | Success: files copied |
| 354 | % lftp bob:h1sp4ssw0rd@172.16.23.1:8021 |
| 355 | lftp bob@172.16.23.1:/> ls |
| 356 | -rw------- 1 bob bob 857648 Jan 01 1970 foobar.jpg |
| 357 | lftp bob@172.16.23.1:/> exit |
| 358 | % lftp alice:h3rp4ssw0rd@172.16.23.1:8021 |
| 359 | lftp alice@172.16.23.1:~> put cliclac.jpg |
| 360 | 857648 bytes transferred |
| 361 | lftp alice@172.16.23.1:~> ls |
| 362 | -rw------- 1 alice alice 857648 Jan 01 1970 foobar.jpg |
| 363 | lftp alice@172.16.23.1:/> exit |
| 364 | % lftp bob:h1sp4ssw0rd@172.16.23.130:8021 |
| 365 | lftp bob@172.16.23.130:~> ls |
| 366 | -rw------- 1 bob bob 857648 Jan 01 1970 foobar.jpg |
| 367 | lftp bob@172.16.23.130:/> exit |
| 368 | % lftp alice:h3rp4ssw0rd@172.16.23.130:8021 |
| 369 | lftp alice@172.16.23.130:~> ls |
| 370 | lftp alice@172.16.23.130:/> exit |
| 371 | }}} |
| 372 | |
| 373 | et voilà ! |
| 374 | |
| 375 | == Configure sFTP access == |
| 376 | |
| 377 | The procedure is slightly the same as for ftp access. You need to have the |
| 378 | ftp.accounts files placed in ~/.tahoe/private/, and we'll consider you already |
| 379 | have the one written in the previous section. |
| 380 | |
| 381 | /!\ As for FTP access, you need to have twisted version >=10.1.0 installed |
| 382 | before trying to use tahoe's sFTP feature. |
| 383 | |
| 384 | {{{ |
| 385 | % ssh-keygen -f ~/.tahoe/private/ssh_host_rsa_key |
| 386 | Generating public/private rsa key pair. |
| 387 | Enter passphrase (empty for no passphrase): |
| 388 | Enter same passphrase again: |
| 389 | Your identification has been saved in /home/guyzmo/.tahoe/private/ssh_host_rsa_key. |
| 390 | Your public key has been saved in /home/guyzmo/.tahoe/private/ssh_host_rsa_key.pub. |
| 391 | The key fingerprint is: |
| 392 | 3f:a6:c7:38:19:3a:19:3c:ee:77:d5:5a:3d:2f:86:6c guyzmo@vm2 |
| 393 | The key's randomart image is: |
| 394 | +--[ RSA 2048]----+ |
| 395 | | | |
| 396 | | | |
| 397 | | | |
| 398 | | | |
| 399 | | . S . . | |
| 400 | | + .. . o..| |
| 401 | | . = =+o + o| |
| 402 | | = =o+.E o .| |
| 403 | | ..o.+ . . . | |
| 404 | +-----------------+ |
| 405 | % vim ~/.tahoe/tahoe.cfg |
| 406 | -------------8<----------------------8<---------------- |
| 407 | 27 [sftpd] |
| 408 | 28 enabled = true |
| 409 | 29 port = 8022 |
| 410 | 30 host_pubkey_file = private/ssh_host_rsa_key.pub |
| 411 | 31 host_privkey_file = private/ssh_host_rsa_key |
| 412 | 32 accounts.file = private/ftp.accounts |
| 413 | ------------->8---------------------->8---------------- |
| 414 | % tahoe restart |
| 415 | STOPPING /home/guyzmo/.tahoe |
| 416 | process 3101 is dead |
| 417 | STARTING /home/guyzmo/.tahoe |
| 418 | client node probably started |
| 419 | % mkdir ./mnt |
| 420 | % mount_sshfs -p 8022 bob@172.16.23.1:/ ./mnt/ |
| 421 | % ls ./mnt/ |
| 422 | total 1676 |
| 423 | -rw-rw-rw- luser staff 857648 Aug 4 17:37 foobar.jpg |
| 424 | }}} |
| 425 | |
| 426 | Now, you have a 3 nodes grid running with 1 introducer, 2 storage servers |
| 427 | and 4 ways to access your files : a WUI, a CLI and accounts for FTP/sFTP |
| 428 | access. |