Move a highly available ArcGIS Data Store to another set of machines

If you need to replace the machines in a highly available ArcGIS Data Store, you can move the installation to a new set of machines provided it is the only type of ArcGIS Data Store running on that set of machines. You might do this if you want the data store to run on newer, faster, or larger-capacity machines.

Note:

In the steps below, it is assumed that there is only one type of data store on the machines you're replacing.

Before you proceed with any of the steps below, create a full backup of the ArcGIS Data Store type to be moved.

The first set of instructions below can be used to move any type of highly available ArcGIS Data Store, but this workflow requires the data store to be inaccessible for the longest amount of time. Options that result in less downtime are available for moving a relational data store or moving a spatiotemporal big data store.

Move any highly available ArcGIS Data Store type to new machines

You can use backup and restore to move any type of highly available ArcGIS Data Store to a new set of machines.

When you use this workflow, the data and, therefore, the web layers that use it are unavailable until you finish moving the installation.

To restore a highly available ArcGIS Data Store to a new set of machines, complete the following steps:

1. Open a command shell on one of the ArcGIS Data Store machines.
2. Run the backupdatastore utility.

   See the utility reference for syntax and rules to run the backupdatastore utility for each data store type.

3. Unregister the data store from the hosting server by doing one of the following based on the type of data store:
   • Relational data store—Run the removemachine utility to remove the standby machine from the relational data store; then run the unregisterdatastore utility to unregister the primary machine from the hosting server.
   • Object store—Run the unregisterdatastore utility to unregister the entire object store cluster from the hosting server, and restart the hosting server.
   • Spatiotemporal big data store—Run the describedatastore utility to determine which machines in the cluster are acting as a coordinator; then run the removemachine utility to remove all but the last coordinator machine from the cluster. The removemachine utility must be run once for each machine. Lastly, run the unregisterdatastore utility to unregister the final machine from the hosting server.
   • Graph store—Run the unregisterdatastore utility to unregister the entire graph store cluster from the hosting server.
4. After the ArcGIS Data Store is unregistered from the hosting server, install ArcGIS Data Store on the new machines.

   Install the same type of data store that you are moving.

   Note:

   Install the software only; do not proceed with configuring the data store.

5. Run the restoredatastore utility to restore the data store to the first new machine using the backup you created in step 3.

How you restore depends on the type of data store.

6. Restore the relational data store by doing the following:
   1. Run the restoredatastore utility on the new primary machine using the following syntax. Replace values in brackets with information specific to your installation.

      restoredatastore --store relational --target <name of backup file>
      --source-loc <backup file location> --data-dir <ArcGIS Data Store content directory location> 
      --server-url <URL of the hosting server in the format https://server.example.com:6443> 
      --server-admin <user name of the hosting server site administrator> 
      --server-password <password of the hosting server site administrator>

   2. Run the configuredatastore utility on the standby machine to register it with the hosting server.

      Data replicates from the primary to the standby.

7. Restore the object store, spatiotemporal big data store, or graph store, by doing the following:
   1. Run the restoredatastore utility on one machine in the cluster but do not load the data.

      Use the syntax below. Replace values in brackets with information specific to your installation.

      restoredatastore --store <data store type> --target <name of backup file>
      --source-loc <backup file location> --data-dir <ArcGIS Data Store content directory location> 
      --server-url <URL of the hosting server in the format https://server.example.com:6443> 
      --server-admin <user name of the hosting server site administrator> 
      --server-password <password of the hosting server site administrator> --mode cluster --loaddata false

   2. On each of the other machines in the cluster, run the configuredatastore utility to complete their configuration.
   3. On the first machine where you restored the data store, run the restoredatastore utility again, but this time, load the data.

      Use the syntax below. Replace values in brackets with information specific to your installation.

      restoredatastore --store <data store type> --target <name of backup file>
      --source-loc <backup file location> --data-dir <ArcGIS Data Store content directory location> 
      --server-url <URL of the hosting server in the format https://server.example.com:6443> 
      --server-admin <user name of the hosting server site administrator> 
      --server-password <password of the hosting server site administrator> --mode cluster --loaddata true

      Data is distributed across machines in the cluster.

8. Test the hosted layers that access the restored data store to ensure that they function.

   If the restoration is successful and the hosted layers function as expected, you can uninstall ArcGIS Data Store from the retired machines.

   If there are problems with the restored data store, retain the installation on the old machines and contact Esri Technical Support or your Esri distributor to help you remove the failed data store and register the old installation with the hosting server.

Move a primary-standby relational data store to two new machines

You can replace both machines in a primary-standby relational data store by replacing the standby machine, promoting it to be the primary machine, and replacing the demoted machine.

When you use this method to move the relational data store to two new machines, hosted feature layers are unavailable for a few seconds while the standby machine is promoted to be the primary machine.

To replace both machines in a relational data store that is running in primary-standby mode, complete the following steps:

1. Install ArcGIS Data Store on the two new machines but do not proceed with configuration yet.
2. Run the removemachine utility to remove the existing standby machine from the data store.
3. Add one new machine as a standby to the data store.

   Depending on the amount of data in the relational data store, it may take several minutes for data to replicate to the standby machine.

4. When all web layer data from the primary machine has been replicated to the standby machine, use the makePrimary command in the ArcGIS Server REST API site to promote the standby to be the primary machine.

   When the new primary machine is online, the old primary will become a standby machine.

5. Access the hosted layers in the ArcGIS Enterprise organization to ensure that they still function properly.
6. Run the removemachine utility to remove the standby machine (the old primary machine that was demoted to be the standby machine) from the data store.
7. Uninstall ArcGIS Data Store from the machine you removed from the data store.
8. Add the second new machine as a standby to the data store.

Move machines in a multimachine spatiotemporal big data store

To replace more than one machine in a multimachine spatiotemporal big data store, add a new machine for each machine you want to replace. When the spatiotemporal big data store finishes replicating data to the new machines, remove the machines you want to retire, one at a time.

Note:

You must wait for the spatiotemporal big data store to rebalance content across the machines before you remove the next machine.

This machine replacement workflow does not result in any downtime for the hosted spatiotemporal feature layers.

To replace more than one machine in a multimachine spatiotemporal big data store, complete the following step:

1. Install ArcGIS Data Store on each new machine.
2. Add the new machine or machines to the existing spatiotemporal big data store, one at a time.
3. When the spatiotemporal big data store finishes replicating data to the new machines, test the web layers in the ArcGIS Enterprise organization to ensure that they're functioning.

   You can use the validate function in the ArcGIS Server Administrator Directory to determine if the data is replicated. If clusterStatus is set to green, the data is replicated.

4. Remove one machine from the spatiotemporal big data store.

   You can use the removemachine utility or the remove command in the ArcGIS Server REST API site.

5. When the data has been rebalanced among the remaining machines, you can remove the next machine if applicable.
6. Uninstall ArcGIS Data Store from the machine you removed from the data store.