John Du wrote:
Rodrigo Costa wrote:
Buchan,
I made exactly what you said even using the -q flag in the slapadd
command. So in summary I did :
1) Load the master DB using LDIF file through slapdd(-q flag and
DB_CACHEZIZE to 1GB);
2) Load the slave DB using LDIF(same) file through slapdd(-q flag and
DB_CACHEZIZE to 1GB);
3)Then have the slapd.conf files appropriately configured
4)Start master and then after some minutes start slave.
So the slave is not starting blank but exactly loaded and in terms of
data there isn't any difference between provider(master) and
consumer(slave). I just see when the consumer slapd process start at
slave machine(slapd started with -d 256 flag) the connection from the
consumer to the provider slapd process.
Then all behavior explained start to happen. Please see the
provider and
consumer configuration file attached.
Something appears not being following the expect behavior. Also the
memory consumed in the consumer is growing too fast and doesn't appear
to really follow the caches directives.
Regards,
Rodrigo.
I do not know what attrs="*,+" is supposed to mean. But with
OpenLDAP2.4.11, if you do a search with attrs="*,+" as the attributes
to search, the search will not return any attributes. Your slave
database will never be synchronized with the master. I'd comment out
that line and try a sync again to determine if this is the cause.
I used an LDIF file to load 4 OpenLDAP servers and the synchronization
works perfectly among the 4 servers after the LDIF file was loaded to
all of them.
Run the following on the command line and you will see what you are
trying to replicate.
ldapsearch -x -h 10.142.15.41 -b "ou=INDEXES,o=domain,c=fr" -D
="cn=admin,ou=INDEXES,o=domain,c=fr" -w secret * "*,+"
The above query is what your specified in your slapd_slave file.