How to compare O/S compute nodes to check configuration across clusters.

There is a node getting evicted and we were asked check the setting on the server what can we see?

This is more of a unix team task well but we collaborate always so here are the things you can compare.

more /etc/sysctl.conf                                         nodea/b/c of oracle RAC
kernel.shmmax = 270471206912                         270,471,206,912
kernel.shmall = 52826408                                  52,826,408
fs.aio-max-nr = 1048576                                    1,048,576
fs.file-max = 6815744                                       6,815,744
kernel.shmmni = 4096                                      4,096
kernel.sem = 250 32000 100 200 250 32000 100 200
kernel.sem = 4100 40000 100 128 4100 40000 100 128
net.ipv4.ip_local_port_range = 9000 65500        9000 65500
net.core.rmem_default = 262144 262,144
net.core.rmem_max = 4194304 4,194,304
net.core.wmem_default = 262144 262,144
net.core.wmem_max = 1048576 1,048,576
vm.nr_hugepages=75000 change it. 150,000
vm.hugetlb_shm_group=501                              501
ipcs -ls
max number of arrays = 200                               200
max semaphores per array = 250 250
max semaphores system wide = 32000 32,000
max ops per semop call = 100 100
semaphore max value = 32767 32,767


Change:

From

kernel.sem = 250 32000 100 200

vm.nr_hugepages=50000


To:

kernel.sem = 250 90000 100 200

vm.nr_hugepages=150000
 "


Compare more /etc/security/limits.conf

Above setting are for 96 Core machine. Make sure that you keep that standard in your mind. 

[NODEa.com: ~]$ free -g

              total        used        free      shared  buff/cache   available

Mem:            503         311          27           4         164         185

Swap:            20           0          20