node agent performance and linux interaction

[thesephirot]

in order to optimize the reconciling process we might think of some optimizations:
I will put some suggestions here, especially for systemd, package management, etc.

for versionchecks it might be:

 time rpm -q --queryformat '%{NAME}-%{VERSION}-%{RELEASE}\n' docker-ce kubelet kubectl kubernetes-cni kubeadm
docker-ce-19.03.5-3.el7
kubelet-1.18.8-0
kubectl-1.18.8-0
kubernetes-cni-0.8.6-0
kubeadm-1.18.8-0

real	0m0.054s
user	0m0.036s
sys	0m0.010s

in comparison to:

time sudo yum list installed
eoIP.x86_64                                                                            1.5.0-14.el7                                                                @base
PyYAML.x86_64                                                                           3.10-11.el7                                                                 installed
acl.x86_64                                                                              2.2.51-15.el7                                                               @base
audit.x86_64                                                                            2.8.5-4.el7                                                                 @base
......
real	0m2.945s
user	0m2.184s
sys	0m0.365s

this will relief the yum db significantly

[thesephirot]

package-cleanup --dupes  

shall ONLY run id the yum db is broken, it runs regulary a the moment each intervall

[thesephirot]

yum install $1 $2 $3

shall only run once/or if it is needed. at the moment it runs regulary each intervall

[thesephirot]

done

systemctl daemon-reload

why is the complete systemctl configuration reloaded ?

[thesephirot]

yum cron disable/enable

yum cron should be enabled ONCE, and will handle the package updates then automatically. there is no need to disable the agent each intervall, install packages and enable the agent again