Can someone shed light on this issue? It now appears, that the replication delay is more to the consumer that is the busiest, but in no way short of system resources.




On Fri, Mar 25, 2016 at 2:45 AM, shardulsk <shardulsk@gmail.com> wrote:
We recently upgraded from Centos 5.4 389-ds Version 1.1.2 to Centos 6.7  389-ds version 1.2.11


389-console-1.1.7-1.el6.noarch
389-ds-base-1.2.11.15-48.el6_6.x86_64
389-ds-console-1.2.6-1.el6.noarch
389-ds-base-libs-1.2.11.15-48.el6_6.x86_64
389-admin-console-1.1.8-1.el6.noarch
389-adminutil-1.1.19-1.el6.x86_64
389-admin-1.1.35-1.el6.x86_64


The setup has a single master, hub and 5 replicas. For some reason we are experiencing replication delays of upto 40 secs between hub and replicas. This did not occur in the old setup.  At the time access logs showed an average of  1000 MOD operations per minute.

Some of our configured parameters:

nsslapd-maxdescriptors: 16384
nsslapd-max-filter-nest-level: 40
nsslapd-timelimit: 7200
nsslapd-sizelimit: 10000000
nsslapd-reservedescriptors: 92
nsslapd-maxthreadsperconn: 10
nsslapd-threadnumber: 120
nsslapd-dbcachesize: 4000000000
nsslapd-cachememsize: 20000000000


The systems resources on the Hub (CPU/memory/disk) look fine, so it must be 389-ds resources either on the hub or the replicas that must be causing the delay.  Where should I be looking?


~Shardul.