| 1 | To set up a new LDAP server: |
|---|
| 2 | |
|---|
| 3 | - Install the RPM 389-ds-base with yum |
|---|
| 4 | root# yum install -y 389-ds-base |
|---|
| 5 | - We want to run the directory server as its own user, so create fedora-ds |
|---|
| 6 | root# env NSS_NONLOCAL_IGNORE=1 useradd -r -d /var/lib/dirsrv fedora-ds |
|---|
| 7 | - root# yum install -y policycoreutils-python |
|---|
| 8 | - Temporarily move away the existing slapd-scripts folder |
|---|
| 9 | root# mv /etc/dirsrv/slapd-scripts{,.bak} |
|---|
| 10 | - root# /usr/sbin/setup-ds.pl |
|---|
| 11 | - Choose a typical install |
|---|
| 12 | - Tell it to use the fedora-ds user and group |
|---|
| 13 | - Directory server identifier: scripts |
|---|
| 14 | Needed to remove this from the config file first |
|---|
| 15 | - Suffix: dc=scripts,dc=mit,dc=edu |
|---|
| 16 | - Input directory manager password |
|---|
| 17 | (this can be found in ~/.ldapvirc) |
|---|
| 18 | - Move the schema back |
|---|
| 19 | root# cp -R /etc/dirsrv/slapd-scripts.bak/{.svn,*} /etc/dirsrv/slapd-scripts |
|---|
| 20 | root# rm -Rf /etc/dirsrv/slapd-scripts.bak |
|---|
| 21 | - yum install ldapvi |
|---|
| 22 | - Check if dirsrv starts: /sbin/service dirsrv start |
|---|
| 23 | then turn it back off: service dirsrv stop |
|---|
| 24 | - Apply the following configuration changes. If you're editing |
|---|
| 25 | dse.ldif, you don't want dirsrv to be on, otherwise it will |
|---|
| 26 | overwrite your changes. [XXX: show how to do these changes with |
|---|
| 27 | dsconf, which is the "blessed" method] |
|---|
| 28 | |
|---|
| 29 | # Inside cn=config. These changes definitely require a restart. |
|---|
| 30 | nsslapd-ldapifilepath: /var/run/dirsrv/slapd-scripts.socket |
|---|
| 31 | nsslapd-ldapilisten: on |
|---|
| 32 | |
|---|
| 33 | # Add these blocks |
|---|
| 34 | |
|---|
| 35 | # mapname, mapping, sasl, config |
|---|
| 36 | # This is the most liberal mapping you can have for SASL: you can |
|---|
| 37 | # basically add authentication for any given GSSAPI mechanism by |
|---|
| 38 | # explicitly creating the UID for that SASL string. |
|---|
| 39 | dn: cn=mapname,cn=mapping,cn=sasl,cn=config |
|---|
| 40 | objectClass: top |
|---|
| 41 | objectClass: nsSaslMapping |
|---|
| 42 | cn: mapname |
|---|
| 43 | nsSaslMapRegexString: \(.*\) |
|---|
| 44 | nsSaslMapBaseDNTemplate: uid=\1,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 45 | nsSaslMapFilterTemplate: (objectClass=posixAccount) |
|---|
| 46 | |
|---|
| 47 | - Put LDAP keytab (ldap/hostname.mit.edu) in /etc/dirsrv/keytab. Make |
|---|
| 48 | sure you chown/chgrp it to be readable by fedora-ds |
|---|
| 49 | - Uncomment and modify in /etc/sysconfig/dirsrv: KRB5_KTNAME=/etc/dirsrv/keytab ; export KRB5_KTNAME |
|---|
| 50 | - chown fedora-ds:fedora-ds /var/run/dirsrv |
|---|
| 51 | - chmod 755 /var/run/dirsrv |
|---|
| 52 | - /sbin/service dirsrv start |
|---|
| 53 | - Use ldapvi -b cn=config to add these indexes (8 of them): |
|---|
| 54 | |
|---|
| 55 | add cn=apacheServerName, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 56 | objectClass: top |
|---|
| 57 | objectClass: nsIndex |
|---|
| 58 | cn: apacheServerName |
|---|
| 59 | nsSystemIndex: false |
|---|
| 60 | nsIndexType: eq |
|---|
| 61 | nsIndexType: pres |
|---|
| 62 | |
|---|
| 63 | add cn=apacheServerAlias, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 64 | objectClass: top |
|---|
| 65 | objectClass: nsIndex |
|---|
| 66 | cn: apacheServerAlias |
|---|
| 67 | nsSystemIndex: false |
|---|
| 68 | nsIndexType: eq |
|---|
| 69 | nsIndexType: pres |
|---|
| 70 | |
|---|
| 71 | add cn=scriptsVhostName, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 72 | objectClass: top |
|---|
| 73 | objectClass: nsIndex |
|---|
| 74 | cn: scriptsVhostName |
|---|
| 75 | nsSystemIndex: false |
|---|
| 76 | nsIndexType: eq |
|---|
| 77 | nsIndexType: pres |
|---|
| 78 | |
|---|
| 79 | add cn=scriptsVhostAlias, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 80 | objectClass: top |
|---|
| 81 | objectClass: nsIndex |
|---|
| 82 | cn: scriptsVhostAlias |
|---|
| 83 | nsSystemIndex: false |
|---|
| 84 | nsIndexType: eq |
|---|
| 85 | nsIndexType: pres |
|---|
| 86 | |
|---|
| 87 | add cn=scriptsVhostAccount, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 88 | objectClass: top |
|---|
| 89 | objectClass: nsIndex |
|---|
| 90 | cn: scriptsVhostAccount |
|---|
| 91 | nsSystemIndex: false |
|---|
| 92 | nsIndexType: eq |
|---|
| 93 | nsIndexType: pres |
|---|
| 94 | |
|---|
| 95 | add cn=memberuid, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 96 | objectClass: top |
|---|
| 97 | objectClass: nsIndex |
|---|
| 98 | cn: memberuid |
|---|
| 99 | nsSystemIndex: false |
|---|
| 100 | nsIndexType: eq |
|---|
| 101 | nsIndexType: pres |
|---|
| 102 | |
|---|
| 103 | add cn=uidnumber, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 104 | objectClass: top |
|---|
| 105 | objectClass: nsIndex |
|---|
| 106 | cn: uidnumber |
|---|
| 107 | nsSystemIndex: false |
|---|
| 108 | nsIndexType: eq |
|---|
| 109 | nsIndexType: pres |
|---|
| 110 | |
|---|
| 111 | add cn=gidnumber, cn=index, cn=userRoot, cn=ldbm database, cn=plugins, cn=config |
|---|
| 112 | objectClass: top |
|---|
| 113 | objectClass: nsIndex |
|---|
| 114 | cn: gidnumber |
|---|
| 115 | nsSystemIndex: false |
|---|
| 116 | nsIndexType: eq |
|---|
| 117 | nsIndexType: pres |
|---|
| 118 | |
|---|
| 119 | - Build the indexes for all the fields: |
|---|
| 120 | |
|---|
| 121 | /usr/lib64/dirsrv/slapd-scripts/db2index.pl -D "cn=Directory Manager" -j /etc/signup-ldap-pw -n userRoot |
|---|
| 122 | |
|---|
| 123 | (/etc/signup-ldap-pw is the LDAP root password, make sure it's |
|---|
| 124 | chmodded correctly and chowned to signup. Also, make sure it doesn't |
|---|
| 125 | have a trailing newline!) |
|---|
| 126 | |
|---|
| 127 | - Watch for the indexing operations to finish with this command: |
|---|
| 128 | |
|---|
| 129 | ldapsearch -x -y /etc/signup-ldap-pw -D 'cn=Directory Manager' -b cn=tasks,cn=config |
|---|
| 130 | |
|---|
| 131 | (look for nktaskstatus) |
|---|
| 132 | |
|---|
| 133 | - Set up replication. |
|---|
| 134 | |
|---|
| 135 | We used to tell people to go execute |
|---|
| 136 | http://directory.fedoraproject.org/sources/contrib/mmr.pl manually |
|---|
| 137 | (manually because that script assumes only two masters and we have |
|---|
| 138 | every one of our servers set up as a master.) However, those |
|---|
| 139 | instructions are inaccurate, because we use GSSAPI, not SSL and |
|---|
| 140 | because the initializing procedure is actually prone to a race |
|---|
| 141 | condition. Here are some better instructions. |
|---|
| 142 | |
|---|
| 143 | LDAP replication is based around producers and consumers. Producers |
|---|
| 144 | push changes in LDAP to consumers: these arrangements are called |
|---|
| 145 | "replication agreements" and the producer will hold a |
|---|
| 146 | nsDS5ReplicationAgreement object that represents this commitment, |
|---|
| 147 | as well as some extra configuration to say who consumers will accept |
|---|
| 148 | replication data from (a nsDS5Replica). |
|---|
| 149 | |
|---|
| 150 | The procedure, at a high level, is this: |
|---|
| 151 | |
|---|
| 152 | 1. Pick an arbitrary existing master. The current server will |
|---|
| 153 | be configured as a slave to that master. Initialize a changelog, |
|---|
| 154 | then request a replication to populate our server with |
|---|
| 155 | information. |
|---|
| 156 | |
|---|
| 157 | M1 <---> M2 ---> S |
|---|
| 158 | |
|---|
| 159 | 2. Configure the new server to be replicated back. |
|---|
| 160 | |
|---|
| 161 | M1 <---> M2 <---> S |
|---|
| 162 | |
|---|
| 163 | 3. Set up the rest of the replication agreements at your leisure. |
|---|
| 164 | |
|---|
| 165 | M1 <---> M2 |
|---|
| 166 | ^ ^ |
|---|
| 167 | | | |
|---|
| 168 | +--> S <--+ |
|---|
| 169 | |
|---|
| 170 | Here's how you do it. |
|---|
| 171 | |
|---|
| 172 | 1. Pull open the replication part of the database. It's fairly empty |
|---|
| 173 | right now. |
|---|
| 174 | |
|---|
| 175 | ldapvi -b cn=\"dc=scripts,dc=mit,dc=edu\",cn=mapping\ tree,cn=config |
|---|
| 176 | |
|---|
| 177 | 2. Configure the server $SLAVE (this server) to accept $MASTER |
|---|
| 178 | replications by adding the following LDAP entries: |
|---|
| 179 | |
|---|
| 180 | add cn=replica, cn="dc=scripts,dc=mit,dc=edu", cn=mapping tree, cn=config |
|---|
| 181 | objectClass: top |
|---|
| 182 | objectClass: nsDS5Replica |
|---|
| 183 | cn: replica |
|---|
| 184 | nsDS5ReplicaId: $REPLICA_ID |
|---|
| 185 | nsDS5ReplicaRoot: dc=scripts,dc=mit,dc=edu |
|---|
| 186 | nsDS5Flags: 1 |
|---|
| 187 | nsDS5ReplicaBindDN: uid=ldap/bees-knees.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 188 | nsDS5ReplicaBindDN: uid=ldap/busy-beaver.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 189 | nsDS5ReplicaBindDN: uid=ldap/cats-whiskers.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 190 | nsDS5ReplicaBindDN: uid=ldap/pancake-bunny.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 191 | nsDS5ReplicaBindDN: uid=ldap/whole-enchilada.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 192 | nsDS5ReplicaBindDN: uid=ldap/real-mccoy.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 193 | nsDS5ReplicaBindDN: uid=ldap/better-mousetrap.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 194 | nsDS5ReplicaBindDN: uid=ldap/old-faithful.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 195 | # ADD SERVERS HERE AS YOU ADD NEW SERVERS |
|---|
| 196 | nsds5ReplicaPurgeDelay: 604800 |
|---|
| 197 | nsds5ReplicaLegacyConsumer: off |
|---|
| 198 | nsDS5ReplicaType: 3 |
|---|
| 199 | |
|---|
| 200 | $REPLICA_ID is the scripts$N number (stella $HOSTNAME to find |
|---|
| 201 | out.) You might wonder why we are binding to all servers; |
|---|
| 202 | weren't we going to replicate from only one server? That is |
|---|
| 203 | correct, however, simply binding won't mean we will receive |
|---|
| 204 | updates; we have to setup the $MASTER to send data $SLAVE. |
|---|
| 205 | |
|---|
| 206 | 3. Although we allowed those uids to bind, that user information |
|---|
| 207 | doesn't exist on $SLAVE yet. So you'll need to create the entry |
|---|
| 208 | for just $MASTER. |
|---|
| 209 | |
|---|
| 210 | add uid=ldap/$MASTER,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 211 | uid: ldap/$MASTER |
|---|
| 212 | objectClass: account |
|---|
| 213 | objectClass: top |
|---|
| 214 | |
|---|
| 215 | 4. Though our $SLAVE will not be making changes to LDAP, we need to |
|---|
| 216 | initialize the changelog because we intend to be able to do this |
|---|
| 217 | later. |
|---|
| 218 | |
|---|
| 219 | add cn=changelog5,cn=config |
|---|
| 220 | objectclass: top |
|---|
| 221 | objectclass: extensibleObject |
|---|
| 222 | cn: changelog5 |
|---|
| 223 | nsslapd-changelogdir: /etc/dirsrv/slapd-scripts/changelogdb |
|---|
| 224 | |
|---|
| 225 | 5. Ok, now go to your $MASTER server that you picked (it should have |
|---|
| 226 | been one of the hosts mentioned in nsDS5ReplicaBindDN) and tell |
|---|
| 227 | it to replicate to $SLAVE. |
|---|
| 228 | |
|---|
| 229 | add cn="GSSAPI Replication to $SLAVE", cn=replica, cn="dc=scripts,dc=mit,dc=edu", cn=mapping tree, cn=config |
|---|
| 230 | objectClass: top |
|---|
| 231 | objectClass: nsDS5ReplicationAgreement |
|---|
| 232 | cn: "GSSAPI Replication to $SLAVE" |
|---|
| 233 | cn: GSSAPI Replication to $SLAVE |
|---|
| 234 | nsDS5ReplicaHost: $SLAVE |
|---|
| 235 | nsDS5ReplicaRoot: dc=scripts,dc=mit,dc=edu |
|---|
| 236 | nsDS5ReplicaPort: 389 |
|---|
| 237 | nsDS5ReplicaTransportInfo: LDAP |
|---|
| 238 | nsDS5ReplicaBindDN: |
|---|
| 239 | uid=ldap/$MASTER,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 240 | nsDS5ReplicaBindMethod: SASL/GSSAPI |
|---|
| 241 | nsDS5ReplicaUpdateSchedule: "0000-2359 0123456" |
|---|
| 242 | nsDS5ReplicaTimeout: 120 |
|---|
| 243 | |
|---|
| 244 | 4. Reboot the server `service dirsrv restart`, then run the |
|---|
| 245 | replication. (Don't fold this into the previous step! You might |
|---|
| 246 | nuke your database!) |
|---|
| 247 | |
|---|
| 248 | # under cn="GSSAPI Replication to $SLAVE", cn=replica, cn="dc=scripts,dc=mit,dc=edu", cn=mapping tree, cn=config |
|---|
| 249 | nsDS5BeginReplicaRefresh: start |
|---|
| 250 | |
|---|
| 251 | 5. Check that the replication is running; the status will be stored |
|---|
| 252 | in the object we've been mucking around with. |
|---|
| 253 | |
|---|
| 254 | If it fails with LDAP Error 49, check /var/log/dirsrv on $MASTER |
|---|
| 255 | for more information. It might be because fedora-ds can't read |
|---|
| 256 | /etc/dirsrv/keytab |
|---|
| 257 | |
|---|
| 258 | 6. Replicate in the other direction. On $MASTER, add $SLAVE |
|---|
| 259 | as a nsDS5ReplicaBindDN in cn=replica,cn="dc=scripts,dc=mit,dc=edu",cn=mapping tree,cn=config |
|---|
| 260 | Also, add an account for $SLAVE |
|---|
| 261 | |
|---|
| 262 | add uid=ldap/$SLAVE,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 263 | uid: ldap/$SLAVE |
|---|
| 264 | objectClass: account |
|---|
| 265 | objectClass: top |
|---|
| 266 | |
|---|
| 267 | On $SLAVE, |
|---|
| 268 | |
|---|
| 269 | add cn="GSSAPI Replication to $MASTER", cn=replica, cn="dc=scripts,dc=mit,dc=edu", cn=mapping tree, cn=config |
|---|
| 270 | objectClass: top |
|---|
| 271 | objectClass: nsDS5ReplicationAgreement |
|---|
| 272 | cn: "GSSAPI Replication to $MASTER" |
|---|
| 273 | cn: GSSAPI Replication to $MASTER |
|---|
| 274 | nsDS5ReplicaHost: $MASTER |
|---|
| 275 | nsDS5ReplicaRoot: dc=scripts,dc=mit,dc=edu |
|---|
| 276 | nsDS5ReplicaPort: 389 |
|---|
| 277 | nsDS5ReplicaTransportInfo: LDAP |
|---|
| 278 | nsDS5ReplicaBindDN: uid=ldap/$SLAVE,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 279 | nsDS5ReplicaBindMethod: SASL/GSSAPI |
|---|
| 280 | nsDS5ReplicaUpdateSchedule: "0000-2359 0123456" |
|---|
| 281 | nsDS5ReplicaTimeout: 120 |
|---|
| 282 | |
|---|
| 283 | If you get a really scary internal server error, that might mean you |
|---|
| 284 | forgot to initialize the changelog. Remove the replication |
|---|
| 285 | agreement (you'll need to turn off dirsrv), add the changelog, and |
|---|
| 286 | then try again. |
|---|
| 287 | |
|---|
| 288 | Troubleshooting |
|---|
| 289 | =============== |
|---|
| 290 | |
|---|
| 291 | LDAP multimaster replication can fail in a number of colorful ways. |
|---|
| 292 | If the failure is local to a single master, usually you can recover |
|---|
| 293 | by asking another master to refresh that master with: |
|---|
| 294 | |
|---|
| 295 | nsDS5BeginReplicaRefresh: start |
|---|
| 296 | |
|---|
| 297 | In practice, we've also had problems with this technique. Some of them |
|---|
| 298 | include: |
|---|
| 299 | |
|---|
| 300 | * Something like https://bugzilla.redhat.com/show_bug.cgi?id=547503 |
|---|
| 301 | on Fedora 11 ns-slapd, where replication is turned off to do the |
|---|
| 302 | replication, but then it wedges and you need to forcibly kill the |
|---|
| 303 | process. |
|---|
| 304 | |
|---|
| 305 | * Failed LDAP authentication because another master attempted to do |
|---|
| 306 | an incremental update. |
|---|
| 307 | |
|---|
| 308 | * Repropagation of the error because the corrupt master thinks it still |
|---|
| 309 | should push updates. |
|---|
| 310 | |
|---|
| 311 | So the extremely safe method to bring up a crashed master is as follows: |
|---|
| 312 | |
|---|
| 313 | 1. Disable all incoming and outgoing replication agreements by editing |
|---|
| 314 | /etc/dirsrv/slapd-scripts/dse.ldif. You'll need to munge: |
|---|
| 315 | |
|---|
| 316 | nsDS5ReplicaBindDN in cn=replica,cn=dc\3Dscripts\2Cdc\3Dmit\2Cdc\3Dedu,cn=mapping tree,cn=config |
|---|
| 317 | |
|---|
| 318 | and all of the push agreements. Deleting them outright works, but |
|---|
| 319 | means you'll have to reconstruct all of the agreements from scratch. |
|---|
| 320 | |
|---|
| 321 | 2. Bring up the server. |
|---|
| 322 | |
|---|
| 323 | 3. Accept incoming replication data from a single server. |
|---|
| 324 | |
|---|
| 325 | 4. Initiate a full update from that server. |
|---|
| 326 | |
|---|
| 327 | 5. Finish setting up replication as described above. |
|---|
| 328 | |
|---|
| 329 | If your database gets extremely fucked, other servers may not be able |
|---|
| 330 | to authenticate because your authentication information has gone missing. |
|---|
| 331 | In that case, the minimal set of entries you need is: |
|---|
| 332 | |
|---|
| 333 | add dc=scripts,dc=mit,dc=edu |
|---|
| 334 | objectClass: top |
|---|
| 335 | objectClass: domain |
|---|
| 336 | dc: scripts |
|---|
| 337 | |
|---|
| 338 | add ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 339 | objectClass: top |
|---|
| 340 | objectClass: organizationalunit |
|---|
| 341 | ou: People |
|---|
| 342 | |
|---|
| 343 | add uid=ldap/real-mccoy.mit.edu,ou=People,dc=scripts,dc=mit,dc=edu |
|---|
| 344 | objectClass: account |
|---|
| 345 | objectClass: top |
|---|
| 346 | uid: ldap/real-mccoy.mit.edu |
|---|