Premium HA - Rebuild data center

Premium High Availability

Dynatrace Premium High Availability (Premium HA) is a self-contained, out-of-the-box solution that provides near-zero downtime and allows monitoring to continue without data loss in failover scenarios. This solution requires additional licensing for your deployment.

In this recovery procedure, the following terms are designated as follows:

Source-DC - Source (surviving) data center where the Managed cluster is located.
Target-DC - Target (lost) data center designated for recovery.
seed node - Any node within Source-DC that will be used for performing the installation tasks and distribution of configuration.

The procedure involves migration and replication of Dynatrace Managed components separately so they are prepared for data replication across two data centers. See Overview of Dynatrace Managed components.

Terminate unavailable data center

Remove nodes

Remove lost data center from configuration

Distribute the installer

Prepare cluster data for a replication

Create the data center topology

Open firewall rules

Install second data center nodes

Migrate Cassandra

Migrate Elasticsearch

Migrate the server

Enable the restored data center

Gather information

The commands will use these variables in executing the REST API calls. For this, you will need the following information:

<seed-node-ip> - The IP address of the seed node from Source-DC.
This can be any node running in existing data center that will be used for performing the installation tasks and distribution of configuration.
<nodes-ips> - The list of IPV4 addresses of new nodes in Target-DC.
Example: "176.16.0.5", "176.16.0.6", "176.16.0.7"
<api-token> - A valid Cluster API token (ServiceProviderAPI scope is required).
You can generate it in the Managed Cluster Management Console. See Cluster API - Authentication.
<dynatrace-directory> - The directory where Dynatrace Managed is installed on the seed node.
The default Dynatrace Managed installation directory is /opt/dynatrace-managed

<datacenter-1> - The Source-DC name must be the same as the Cassandra DC name.
The default Cassandra DC name is datacenter1.

Get the DC name.

To get the DC name, execute this command on the seed node before starting migration:

sudo <dynatrace-directory>/utils/cassandra-nodetool.sh status

You will get a response that includes the Source-DC name. Example for a DC named datacenter1:

plaintext
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address        Load       Tokens       Owns (effective)  Host ID                               Rack
UN  10.176.42.20   65.54 GB   256          100.0%            f053dd8d-ecf3-7834-b099-68542439817b  rack1
UN  10.176.42.244  65.47 GB   256          100.0%            2aa7e790-a423-9273-88f9-45bcd158dd6e  rack1
UN  10.176.42.168  65.47 GB   256          100.0%            48543bca-41f5-26d3-b2fd-6cfdf5c0f3b2  rack1

<datacenter-2> - The Target-DC name must remain unchanged. Example: dc-us-east-2.

Lost data center name

You must use the same name of the lost data center during the recovery to Target-DC.

Set variables

Set the following environment variables on seed node in Source-DC and EACH NODE in Target-DC:

shell
SEED_IP=<seed-ip>
DT_DIR=<dynatrace-directory>
NODES_IPS=echo '[<nodes-ips]'
API_TOKEN=<api-token>
SDC_NAME=<datacenter-1>
TDC_NAME=<datacenter-2>

For example:

shell
SEED_IP=10.176.37.201
DT_DIR=/opt/dynatrace-managed
NODES_IPS=echo '["10.176.37.218", "10.176.37.227", "10.176.37.120"]'
API_TOKEN=R_SZOpV4RTOmjr9fFmK4x
SDC_NAME=datacenter1
TDC_NAME=dc-us-east-2

Check for custom settings

If your Cassandra or Elasticsearch cluster is configured with custom.settings that enable rack-awareness, contact Dynatrace ONE support to apply these custom settings before proceeding with Target-DC recovery.

To check whether custom settings are applied, execute on seed node:

plaintext
ls $DT_DIR/installer/custom.settings

If the custom.settings file exists, you are using custom settings.

Recovery

To restore a cluster in a Target-DC, follow this procedure.

API return codes

Each of the REST API calls will return the HTTP code. Go to the next step only when the returned code is 200. Expect the following return codes:

200 - Go to the next step, current step was executed successfully.
207 - The request is being processed, repeat a step after few minutes.
40x - Revise your request path and arguments and repeat the request.
5xx - Contact support.

Terminate unavailable data center

Stop all Dynatrace Managed required services in the recommended order. See Start/stop/restart a cluster.

Remove nodes

Go to the Dynatrace Managed Cluster Management Console.
For each node in unavailable data center, go to the Node details page and remove the node.

For more information on other ways to remove a node, see Remove a cluster node.

Remove lost data center from configuration

Execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X POST https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/lostDatacenterCleanUp?Api-Token=$API_TOKEN -H  "accept: application/json" -H  "Content-Type: application/json"

If the status code is not 200 and the response does not suggest next steps, contact Dynatrace ONE.

Distribute the installer

In this step, you will copy the node installer to every node in Target-DC.

Log into your Dynatrace Managed Cluster Management Console.
In the Dynatrace menu, go to Home for the Dynatrace Managed deployment status page.
Click Add new cluster node.
Copy the wget command line from the Run this command on the target host text box.

Do not run the installer script
The Run this installer script with root rights text box contains a command for the installation script. Ignore this command; do not execute the provided script.
Paste and execute only the wget command line into your terminal window.

Prepare cluster data for a replication

In this step, you will prepare data indexes for replication.

Prepare cluster data

Execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X POST https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/clusterReplicationPreparation?Api-Token=$API_TOKEN

If the status code is not 200 and the response does not suggest next steps, contact Dynatrace ONE.

Check cluster preparation status

Execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/clusterReplicationPreparation?Api-Token=$API_TOKEN -H  "accept: application/json"

If the status code from this call is not 200, try again after a few minutes.

Create the data center topology

In this step, you will create configuration that maintains which nodes belong to which data center.

Execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X POST -d "{\"newDatacenterName\" : \"$TDC_NAME\", \"nodesIp\" :$NODES_IPS}" https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/datacenterTopology?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code is not 200 and the response does not suggest next steps, contact Dynatrace ONE.

Open firewall rules

In this step, you will add firewall rules that open ports to traffic from the Target-DC nodes.

Open ports

To open ports to traffic from the new Target-DC nodes, execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X POST -d "$NODES_IPS" https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/clusterNodes/currentDc?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If successful, status code will be 200 and the response body will contain a request ID you need to check the firewall rules status.

If the status code is not 200 and the response does not suggest next steps, contact Dynatrace ONE.

Verify firewall rules

Set the request ID environment variable on seed node only. The request ID is from the response in the previous API call.

shell
REQ_ID=<topology-configuration-request-id>

To check the firewall rules status, execute the following cluster API call only on the seed node:

plaintext
curl -ikS https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/clusterNodes/currentDc/$REQ_ID?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code from this call is not 200, try again after a few minutes.

Install second data center nodes

In this step, you will install manged nodes on all hosts within Target-DC and once completed, you will check for the presence of a Nodekeeper service. This will indicate if all nodes were successfully installed in Target-DC.

Install nodes in Target-DC

Execute the following command on every node in Target-DC. Follow the on-screen instructions as this will be a typical node installation.

plaintext
sudo /bin/sh ./managed-installer.sh --install-new-dc --premium-ha on --datacenter $TDC_NAME --seed-auth $API_TOKEN

This operation should take about 3-5 minutes and the expected result should be similar to this:

plaintext
Installation in new data center completed successfully after 2 minutes 51 seconds.

Check Nodekeeper in Target-DC

Execute the following cluster API call only on the seed node when all nodes in Target-DC finish installing:

plaintext
curl -ikS https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/nodekeeper/healthCheck?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code is not 200, try again after a few minutes.

Migrate Cassandra

In this step, you will reconfigure Cassandra in Source-DC and Target-DC for cross data center replication, trigger data synchronization, rebuild Cassandra data and verify Cassandra state.

It may take minutes to hours depending on your metric storage size.

Migration of Cassandra in Target-DC

In this step, you will reconfigure Cassandra for cross data center replication and trigger data synchronization.

Migrate Cassandra

To start migration of Cassandra in the Target-DC data center, execute the following cluster API call only on the seed node:
```
plaintext
curl -ikS -X POST https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/cassandra/newDc?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"
```
If successful, status code will be 200 and the response body will contain a request ID which you need to check migration status. Set the request ID environment variable only on the seed node. The request ID is from the response in the previous API call.
```
shell
REQ_ID=<migration-new-datacenter-request-id>
```
If status code is not 200 and response does not suggest next steps, contact Dynatrace ONE.

Check migration status

To check the migration status, execute the following cluster API call only on the seed node:
```
plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/cassandra/newDc/$REQ_ID?Api-Token=$API_TOKEN -H  "accept: application/json" -H  "Content-Type: application/json"
```
If the status code is not 200, try again after few minutes.
Rebuild Cassandra data

In this step, you will rebuild Cassandra, verify the progress by checking the status. Depending on the size of your Cassandra database, can take several hours.

Rebuild data

To rebuild Cassandra, run the following command on each new Target-DC node successively. Use the nohup command to prevent interruption of script execution (such as session disconnect) during important operations.
```
plaintext
sudo nohup $DT_DIR/utils/cassandra-nodetool.sh rebuild -- $SDC_NAME &
```
Verify progress and status

To verify the progress and status, execute the following cluster API call only on the seed node:
```
plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/migration/cassandra/rebuildStatus?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"
```
If the status code is not 200, try again after approximately 15 minutes. Remember that the rebuild process can be time-consuming.

Verify Cassandra state

To verify Cassandra cluster state, execute the cassandra-nodetool.sh with the status parameter only on the seed node:

plaintext
sudo $DT_DIR/utils/cassandra-nodetool.sh status

The result should look similar to this:

plaintext
Datacenter: dc1
===============
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address        Load       Tokens       Owns (effective)  Host ID                               Rack
UN  10.176.41.167  18.82 GB   256          100.0%            3af25127-4f99-4f43-afc3-216d7a2c10f8  rack1
UN  10.176.41.154  19.44 GB   256          100.0%            5a618559-3a73-42ec-83f0-32d28e08beec  rack1
UN  10.176.41.43   19.58 GB   256          100.0%            191f3b30-949a-4cf2-b620-68a40eebf31e  rack1
Datacenter: dc2
===============
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address        Load       Tokens       Owns (effective)  Host ID                               Rack
UN  10.176.42.54   19.18 GB   256          100.0%            852ce236-a430-400a-92a6-daeed99acf68  rack1
UN  10.176.42.104  19.12 GB   256          100.0%            84479219-b64d-442c-a807-a832db9aae18  rack1
UN  10.176.42.234  19.4 GB    256          100.0%            507b377c-5bfc-4667-b251-a9b7c453ed22  rack1

The Load value should not differ significantly between the nodes and Status should be UN on all nodes.

Migrate Elasticsearch

In this step, you will migrate Elasticsearch to the Target-DC data center, verify the configuration and data migration. This step may take minutes or hours depending on your Elasticsearch storage.

Migrate Elasticsearch to Target-DC

Start Elasticsearch. Execute the following command successively on every node in Target-DC only:

plaintext
sudo $DT_DIR/launcher/elasticsearch.sh start

To start migration of Elasticsearch to the Target-DC data center, execute the following cluster API call only on the seed node:

curl -ikS -X POST https://$SEED_IP/api/v1.0/onpremise/multiDc/restore/elasticsearch/recover?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If successful, the status code will be 200.

Verify progress and status

To check the migration status of Elasticsearch, execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/restore/elasticsearch/recover?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code is not 200, try again after a few minutes.

Verify data migration

To verify migration Elasticsearch data migration, execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/restore/elasticsearch/recover/indexMigrationStatus?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code is not 200, try again after a few minutes.

Migrate the server

In this step, you will migrate the server, start ActiveGate, and start NGINX in the Target-DC data center.

Migrate server

Launch the Managed cluster in the Target-DC by executing the following cluster API call only on the seed node:

plaintext
curl -ikS -X POST https://$SEED_IP/api/v1.0/onpremise/multiDc/restore/server/recovery?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If successful, status code will be 200 and the response body will contain a request ID which you need to check cluster readiness. Set the request ID environment variable only on the seed node. The request ID is from the response in the previous API call.

shell
REQ_ID=<migration-server-request-id>

If status code is not 200 and response does not suggest next steps, contact Dynatrace ONE.

Check cluster readiness

To check if the cluster is ready, execute the following cluster API call only on the seed node:

plaintext
curl -ikS -X GET https://$SEED_IP/api/v1.0/onpremise/multiDc/restore/server/recovery/$REQ_ID?Api-Token=$API_TOKEN -H "accept: application/json" -H "Content-Type: application/json"

If the status code is not 200, try again after a few minutes.

Enable the restored data center

Enable OneAgent traffic.
For details, see Enable/Disable a cluster node.
Enable backup in one of the data centers. Your backup is disabled after migration.
For details, see Back up and restore a cluster.

Premium HA - Rebuild data center

Gather information

Set variables

Check for custom settings

Recovery

Terminate unavailable data center

Remove nodes

Remove lost data center from configuration

Distribute the installer

Prepare cluster data for a replication

Prepare cluster data

Check cluster preparation status

Create the data center topology

Open firewall rules

Open ports

Verify firewall rules

Install second data center nodes

Install nodes in Target-DC

Check Nodekeeper in Target-DC

Migrate Cassandra

Migration of Cassandra in Target-DC

Migrate Cassandra

Check migration status

Rebuild Cassandra data

Rebuild data

Verify progress and status

Verify Cassandra state

Migrate Elasticsearch

Migrate Elasticsearch to Target-DC

Verify progress and status

Verify data migration

Migrate the server

Migrate server

Check cluster readiness

Enable the restored data center