Steps that may help improve NMS stability
======================================
Increase the nms memory-check trigger threshold to restart the NMS process
e.g. when it reaches 256Mb RAM - if it is set too low, it may never trigger and therefore never restart
NMC> setup trigger memory-check property max_rss_suspect -p 256
Limit the number of starting nms instances to one, which may improve system stability
NMC> setup appliance nms property srvpool_cnt_initial -p 0
Below is reference for default value.
======================================
nmc@node2:/$ show trigger memory-check property | grep max
flags : maintenance on max fail | clear old faults
max_fault_cnt : 5 # Number of times to register the fault condition prior to generating email notification
max_rss_suspect : 128 # Maximum resident set size (RSS) of any process in the system (megabytes) ★
vmstat_aggr_max : 60 # Statistics aggregation period (seconds)
nmc@node2:/$ show appliance nms property | grep srvpool
srvpool_cnt_initial 2 Initial size of the pool of Management Server ★
srvpool_port_range_min 2011 TCP port for the first Management Server in the pool to listen on
srvpool_interface_serialization_timeout 60 Time interval to serialize API calls through a given interface object (provided by a given management server) after any
srvpool_cnt_max 0 Maximum size of the pool of Management Servers
srvpool_affinity_timeout 60 Time interval to retain client => NMS-from-the-pool relationship (connection affinity)