Steffen Blume wrote:
> Hi,
>
> I have tried to setup up multi master replication without success. The
> two ldap servers are running fine. Then I execute the mmr.pl script (on b):
> ./mmr.pl --host1 a.domain.local --host2 b.domain.local --bindpw secret
> --host1_id 1 --host2_id 2 --repmanpw secret --base "dc=domain, dc=local"
> --create
>
> --- error log on a ---
> [01/Sep/2010:14:11:39 +0200] NSMMReplicationPlugin -
> agmt="cn="Replication to b.domain.local"" (b:389): Replica has a
> different generation ID than the local data.
> [01/Sep/2010:14:11:42 +0200] NSMMReplicationPlugin - Beginning total
> update of replica "agmt="cn="Replication to b.domain.local""
(b:389)".
> [01/Sep/2010:14:11:47 +0200] NSMMReplicationPlugin - Finished total
> update of replica "agmt="cn="Replication to b.domain.local""
(b:389)".
> Sent 1375 entries.
> --------------------
>
> --- error log on b ---
> [01/Sep/2010:14:11:39 +0200] NSMMReplicationPlugin -
> agmt="cn="Replication to a.domain.local"" (a:389): Replica has a
> different generation ID than the local data.
> [01/Sep/2010:14:11:40 +0200] NSMMReplicationPlugin -
> repl_set_mtn_referrals: could not set referrals for replica
> dc=domain,dc=local: 32
> [01/Sep/2010:14:11:40 +0200] NSMMReplicationPlugin -
> multimaster_be_state_change: replica dc=domain,dc=local is going
> offline; disabling replication
> [01/Sep/2010:14:11:41 +0200] - somehow, there are still 200 entries in
> the entry cache. :/
> [01/Sep/2010:14:11:42 +0200] - WARNING: Import is running with
> nsslapd-db-private-import-mem on; No other process is allowed to access
> the database
> [01/Sep/2010:14:11:46 +0200] - import userRoot: Workers finished;
> cleaning up...
> [01/Sep/2010:14:11:46 +0200] - import userRoot: Workers cleaned up.
> [01/Sep/2010:14:11:46 +0200] - import userRoot: Indexing complete.
> Post-processing...
> [01/Sep/2010:14:11:46 +0200] - import userRoot: Flushing caches...
> [01/Sep/2010:14:11:46 +0200] - import userRoot: Closing files...
> [01/Sep/2010:14:11:46 +0200] - somehow, there are still 200 entries in
> the entry cache. :/
> [01/Sep/2010:14:11:47 +0200] - import userRoot: Import complete.
> Processed 1375 entries in 5 seconds. (275.00 entries/sec)
> [01/Sep/2010:14:11:47 +0200] NSMMReplicationPlugin -
> multimaster_be_state_change: replica dc=domain,dc=local is coming
> online; enabling replication
> [01/Sep/2010:14:11:47 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
>
>
This means the RUV entry or some other MMR state information was left
over from a previous configuration attempt. Err=68 is Already Exists -
the entry already exists.
Since this fails, nothing else is going to work.
> [01/Sep/2010:14:11:47 +0200] NSMMReplicationPlugin -
> replica_enable_replication: reloading ruv failed
> [01/Sep/2010:14:11:49 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
> [01/Sep/2010:14:12:19 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
> [01/Sep/2010:14:12:49 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
> [01/Sep/2010:14:13:19 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
> [01/Sep/2010:14:13:49 +0200] NSMMReplicationPlugin -
> _replica_configure_ruv: failed to create replica ruv tombstone entry
> (dc=domain, dc=local); LDAP error - 68
> --------------------
>
> So what do the errors "repl_set_mtn_referrals: could not set referrals"
> and "_replica_configure_ruv: failed to create replica ruv tombstone
> entry" mean?
>
> The messages on b stop, when I restart the ldap server. But the
> replication is not working.
>
Since MMR setup failed, no MMR is going to work.
> On the first replication setup not all the
> data was copied. I removed the replication configuration with mmr.pl
>
I think this is the problem. Either mmr.pl does not cleanly remove the
replication configuration, or there is a bug in the server. For
example, see
https://bugzilla.redhat.com/show_bug.cgi?id=624442
> and
> set it up again with same error messages.
> When I change something (in uid=sbl,ou=people,...) on a the error log of
> a shows
> --- error log on a ---
> [01/Sep/2010:14:35:20 +0200] NSMMReplicationPlugin -
> agmt="cn="Replication to b.domain.local"" (b:389): Replica has a
> different generation ID than the local data.
> [01/Sep/2010:14:35:24 +0200] NSMMReplicationPlugin -
> agmt="cn="Replication to b.domain.local"" (b:389): Replica has a
> different generation ID than the local data.
> [01/Sep/2010:14:35:28 +0200] NSMMReplicationPlugin -
> agmt="cn="Replication to b.domain.local"" (b:389): Replica has a
> different generation ID than the local data.
> ...
> --------------------
>
>
This means the consumer was not initialized properly.
> Nothing in error log on b. But in access log:
>
> --- acces log on b ---
> [01/Sep/2010:14:35:20 +0200] conn=0 op=3 SRCH base="ou=People,
> dc=domain, dc=local" scope=1 filter="(objectClass=*)"
attrs="objectClass"
> [01/Sep/2010:14:35:20 +0200] conn=0 op=7 EXT
> oid="2.16.840.1.113730.3.5.3" name="Netscape Replication Start
Session"
> [01/Sep/2010:14:35:20 +0200] conn=0 op=7 RESULT err=0 tag=120 nentries=0
> etime=0
> [01/Sep/2010:14:35:20 +0200] conn=0 op=8 EXT
> oid="2.16.840.1.113730.3.5.5" name="Netscape Replication End
Session"
> [01/Sep/2010:14:35:20 +0200] conn=0 op=8 RESULT err=0 tag=120 nentries=0
> etime=0
> [01/Sep/2010:14:35:20 +0200] conn=0 op=3 RESULT err=0 tag=101
> nentries=100 etime=0 notes=U
> [01/Sep/2010:14:35:20 +0200] conn=0 op=4 SRCH base="ou=People,
> dc=domain, dc=local" scope=1 filter="(objectClass=*)"
attrs="objectClass"
> [01/Sep/2010:14:35:20 +0200] conn=0 op=4 RESULT err=0 tag=101
> nentries=82 etime=0
> [01/Sep/2010:14:35:24 +0200] conn=0 op=10 EXT
> oid="2.16.840.1.113730.3.5.3" name="Netscape Replication Start
Session"
> [01/Sep/2010:14:35:24 +0200] conn=0 op=10 RESULT err=0 tag=120
> nentries=0 etime=0
> [01/Sep/2010:14:35:24 +0200] conn=0 op=11 EXT
> oid="2.16.840.1.113730.3.5.5" name="Netscape Replication End
Session"
> [01/Sep/2010:14:35:24 +0200] conn=0 op=11 RESULT err=0 tag=120
> nentries=0 etime=0
> [01/Sep/2010:14:35:25 +0200] conn=0 op=5 SRCH
> base="uid=sbl,ou=People,dc=domain,dc=local" scope=0
> filter="(objectClass=*)" attrs=ALL
> [01/Sep/2010:14:35:25 +0200] conn=0 op=5 RESULT err=0 tag=101 nentries=1
> etime=0
> [01/Sep/2010:14:35:27 +0200] conn=0 op=12 EXT
> oid="2.16.840.1.113730.3.5.3" name="Netscape Replication Start
Session"
> [01/Sep/2010:14:35:27 +0200] conn=0 op=12 RESULT err=0 tag=120
> nentries=0 etime=0
> [01/Sep/2010:14:35:27 +0200] conn=0 op=13 EXT
> oid="2.16.840.1.113730.3.5.5" name="Netscape Replication End
Session"
> [01/Sep/2010:14:35:27 +0200] conn=0 op=13 RESULT err=0 tag=120
> nentries=0 etime=0
> ...
> --------------------
>
> Both 389 DS versions are 1.2.4. I compiled it myself for OpenSolaris
> (SunOS 5.11 snv_111b)
>
>
Try 1.2.6. There have been many, many bug fixes between 1.2.4 and 1.2.6.
I compiled 1.2.6 today and installed it on A and B. I converted my
database on A to the new format. Then I set up a new clean instance on B
and tried to set up replication. The error log on B gave me some errors
about entries without parents. So I decided to export my whole data
(userRoot) on A into an ldif, set up a new clean instance on A and B,
import the data on A and set up replication using the mmr script again.
And again I got the error message: "failed to create replica ruv
tombstone entry (dc=domain,dc=local); LDAP error - 68" And the initial
copy is not complete!
Replication info is stored below netscapeRoot that I did not ex- or
import, so this cannot be the problem this time, right?
I don't think, the error logs are helpful but here they are for
completeness:
--- errors on A ---
389-Directory/1.2.6 B2010.253.730
a.domain.local:636 (/usr/ldap/etc/dirsrv/slapd-a)
[10/Sep/2010:17:59:34 +0200] - 389-Directory/1.2.6 B2010.253.730 starting up
[10/Sep/2010:17:59:35 +0200] - slapd started. Listening on All
Interfaces port 389 for LDAP requests
[10/Sep/2010:17:59:35 +0200] - Listening on All Interfaces port 636 for
LDAPS requests
[10/Sep/2010:18:01:28 +0200] NSMMReplicationPlugin -
agmt="cn=Replication to b.domain.local" (b:389): Replica has a different
generation ID than the local data.
[10/Sep/2010:18:01:32 +0200] NSMMReplicationPlugin - Beginning total
update of replica "agmt="cn=Replication to b.domain.local" (b:389)".
[10/Sep/2010:18:01:37 +0200] NSMMReplicationPlugin - Finished total
update of replica "agmt="cn=Replication to b.domain.local" (b:389)".
Sent 1375 entries.
--------------------------
--- errors on B ---
389-Directory/1.2.6 B2010.253.730
b.domain.local:636 (/usr/ldap/etc/dirsrv/slapd-b)
[10/Sep/2010:17:58:28 +0200] - 389-Directory/1.2.6 B2010.253.730 starting up
[10/Sep/2010:17:58:29 +0200] - slapd started. Listening on All
Interfaces port 389 for LDAP requests
[10/Sep/2010:17:58:29 +0200] - Listening on All Interfaces port 636 for
LDAPS requests
[10/Sep/2010:18:01:29 +0200] NSMMReplicationPlugin -
agmt="cn=Replication to a.domain.local" (a:389): Replica has a different
generation ID than the local data.
[10/Sep/2010:18:01:30 +0200] NSMMReplicationPlugin -
multimaster_be_state_change: replica dc=domain,dc=local is going
offline; disabling replication
[10/Sep/2010:18:01:31 +0200] - entrycache_clear_int: there are still 6
entries in the entry cache. :/
[10/Sep/2010:18:01:31 +0200] - dncache_clear_int: there are still 6 dn's
in the dn cache. :/
[10/Sep/2010:18:01:31 +0200] - WARNING: Import is running with
nsslapd-db-private-import-mem on; No other process is allowed to access
the database
[10/Sep/2010:18:01:35 +0200] - import userRoot: Workers finished;
cleaning up...
[10/Sep/2010:18:01:36 +0200] - import userRoot: Workers cleaned up.
[10/Sep/2010:18:01:36 +0200] - import userRoot: Indexing complete.
Post-processing...
[10/Sep/2010:18:01:36 +0200] - import userRoot: Flushing caches...
[10/Sep/2010:18:01:36 +0200] - import userRoot: Closing files...
[10/Sep/2010:18:01:36 +0200] - entrycache_clear_int: there are still 17
entries in the entry cache. :/
[10/Sep/2010:18:01:36 +0200] - dncache_clear_int: there are still 1375
dn's in the dn cache. :/
[10/Sep/2010:18:01:36 +0200] - import userRoot: Import complete.
Processed 1375 entries in 5 seconds. (275.00 entries/sec)
[10/Sep/2010:18:01:36 +0200] NSMMReplicationPlugin -
multimaster_be_state_change: replica dc=domain,dc=local is coming
online; enabling replication
[10/Sep/2010:18:01:36 +0200] NSMMReplicationPlugin -
_replica_configure_ruv: failed to create replica ruv tombstone entry
(dc=domain,dc=local); LDAP error - 68
[10/Sep/2010:18:01:36 +0200] NSMMReplicationPlugin -
replica_enable_replication: reloading ruv failed
[10/Sep/2010:18:01:38 +0200] NSMMReplicationPlugin -
_replica_configure_ruv: failed to create replica ruv tombstone entry
(dc=domain,dc=local); LDAP error - 68
[10/Sep/2010:18:02:08 +0200] NSMMReplicationPlugin -
_replica_configure_ruv: failed to create replica ruv tombstone entry
(dc=domain,dc=local); LDAP error - 68
--------------------------
--
Dipl.-Ing. Steffen Blume Institute of Microelectronic Systems
phone : +49-511-762-19605 Leibniz Universität Hannover
fax : +49-511-762-19601 Appelstr. 4, 30167 Hannover, Germany
mail : sblume(a)ims.uni-hannover.de