When using the Elastifile Cloud File System advanced features, like - AsyncDR or Snapshot Backup, the load on EMS pd-standard may reach the pd i/o limit.
To solve the issue it is recommended to move the EMS instance to use PD-SSD.
When instructed, this procedure need to be performed online with Elastifile support.
First we need to populate all variables used during this KB. -
export EMS_INSTANCE_NAME=""
export GCP_PROJECT_ID=""
export GCP_REGION=""
export GCP_ZONE=""
export SNAPSHOT_NAME=$EMS_INSTANCE_NAME"-snapshot"
export DISK_NAME=$EMS_INSTANCE_NAME
EMS_INSTANCE_NAME = Current EMS Instance name in GCP console
GCP_PROJECT_ID = GCP project ID used by the current EMS instance
GCP_REGION = GCP region used by current EMS instance
GCP_ZONE = GCP zone used by current EMS
EMS_INSTANCE_NAME-snapshot = Snapshot name for the current EMS boot disk.
DISK_NAME = Current EMS boot device name, by default same as EMS instance name.
Stop current EMS instance -
From this step till procedure is done, EMS UI & CLI will not be accessible, NFS i/o will continue as usual, without interruption.
gcloud compute instances stop $EMS_INSTANCE_NAME --project $GCP_PROJECT_ID --zone $GCP_ZONE
Snapshot current EMS disk -
gcloud compute disks snapshot $EMS_INSTANCE_NAME --storage-location $GCP_REGION --project $GCP_PROJECT_ID --zone $GCP_ZONE --snapshot-names $SNAPSHOT_NAME
Detach current EMS hdd boot device -
gcloud compute instances detach-disk $EMS_INSTANCE_NAME--disk=$DISK_NAME--zone$GCP_ZONE --project $GCP_PROJECT_ID
Delete current EMS hdd boot disk -
Don't progress to this step until all previous steps finished successfully.
gcloud compute disks delete $DISK_NAME --zone $GCP_ZONE --project $GCP_PROJECT_IDCreate new EMS boot device based on pd-ssd -
gcloud compute disks create $DISK_NAME --size 100Gb --source-snapshot$SNAPSHOT_NAME--type=pd-ssd--zone$GCP_ZONE --project $GCP_PROJECT_ID
Attach new ssd boot device to EMS -
gcloud compute instances attach-disk$EMS_INSTANCE_NAME--disk=$DISK_NAME --boot--zone$GCP_ZONE --project $GCP_PROJECT_ID
gcloud compute instances set-disk-auto-delete $EMS_INSTANCE_NAME --auto-delete --disk=$DISK_NAME --zone=$GCP_ZONE
Start EMS -
gcloud compute instances start $EMS_INSTANCE_NAME --project $GCP_PROJECT_ID --zone $GCP_ZONE
New EMS is ready, connect to the EMS UI CLI, and check is functioning correctly.
- UI - main dashboard work.
- UI - check standard flows used daya by day.
- CLI run -
source /root/elfs_admin;elfs-cli system show --id 1
- CLI run -
ecs-cli system
If all the commands above, return ok and the UI is functioning correctly, then the migration to ssd boot device finished successfully.