Open tdudgeon opened 3 years ago
Seems like the instances can just be resized to c-class instances. Jacob says this is the recommended approach, but best first to create a snapshot just in case.
I did a quick test and it worked fine with an ubuntu VM, but I'm having problems creating centos7 ones ATM so can't test that.
The following nodes have now been removed form the xchem-follow-up cluster: -
xch-ctrl1
xch-etcd1
pulsar-bastion-2
was resized to a c2.small
instance.
rke-etcd-1/2/3
have been resized to c1.large
instances.
pulsar-central-manager.xchem
has been resized to a c2.large
instance.
Only the xch-production-etcd1/2/3 and xch-production-ctrl1/2 instances remain and those will be handled as part of the prod K8S cluster upgrade.
The m class hypervisors are starting to be removed so we must replace the VMs with c class ones. Seems like we have the following ones:
DLS project pulsar-central-manager.xchem (m2.large) pulsar-bastion-2 (m1.small)
xchem-follow-up project xch-ctrl1 (m1.xlarge) xch-etcd1 (m1.large) kubernetes-nfs-fs-media (m1.large) xch-production-etcd1/2/3 (m1.large) xch-production-ctrl1/2 (m1.large) rke-etcd-1/2/3 (m1.large)