Configuring Junos Space to write RRD files to external storage as a data loss prevention measure.

Junos Space network monitoring (openNMS) writes RRD files to hard disk to provide time-series historical graphs.
RRD files cause a large number of small random disk writes usually with a few writes per update.
In some fail-over situations, there may be loss of a few RRD file writes.
This loss of data will not cause complete failure of the graph data but may cause a small time gap in the available data.

As a workaround, Juniper recommends to have these RRD files written on an external drive (NFS server) which will be accessible by both Space nodes. In case of failover, no data is lost as both Junos Space nodes are writing and reading from a central data storage.   This central data storage would only be for openNMS RRD files.

This requires configuration changes for Opennms. 

Below are detailed steps to be done on the NFS server and Space nodes.
You need a linux server which would act as a NFS server.

Do the following on NFS server:

service portmap stop
service nfs stop
groupadd -g 504 opennms
useradd -u 503 -g opennms -G opennms opennms
mkdir /opennmsrrd
chown opennms:opennms /opennmsrrd
vi /etc/exports, add the following line
/opennmsrrd *(rw,sync,all_squash,anonuid=503,anongid=504)
service portmap start
service nfs start


Do the following changes on both Space nodes.

service jmp-watchdog stop
service jmp-opennms stop
mv /var/opennms/rrd /var/opennms/rrd.old
mkdir –p /var/opennms/rrd
chown opennms:opennms /var/opennms/rrd
mount -t nfs -o nolock 10.206.40.38:/opennmsrrd /var/opennms/rrd/
cp –rf /var/opennms/rrd.old/* /var/opennms/rrd/
service jmp-watchdog start
vi /etc/csync2/csync2.cfg
comment out the following 2 lines
#include /var/opennms/rrd/response/;
#include /var/opennms/rrd/snmp/;