I am running a System with 64 Cores and 256 GB RAM as Solaris 10u10 Virtualization Hosts (Solaris Zones).
There are 18 zones running on the System, all with capped-memory and dedicated-cpus (ncpus attribute). The global-zone itself got 24 GB of RAM and 6 Cores left, everything else is given to the zones. The system runs fluid, but sometimes our systemchecks (like nrpe, snmp and even ping) are failing. Maybe there are not enough resources for the global zone?!
Can anybody tell how man ressources has to be left out for the global-zone itself??
I heared you should left 2% of the Hosts ressources for each Zone running on it.... In this example, that would be around 96GB and 23 Cores that has to be left for the global zone. This seems to be really much.
will you please monitor your system load avergaes including zones
can u share the o/p of poolstat,uptime,swap -l,swap allocation&zpool ouput when issue occures