Restoring the system configuration

Use this procedure to restore the system configuration in the following situations: only if the recover system procedure fails or if the data that is stored on the volumes is not required.

Before you begin

Important: Before running a T4 procedure, contact IBM® support for assistance.

This configuration restore procedure is designed to restore information about your configuration, such as volumes, local Metro Mirror information, local Global Mirror information, storage pools, and nodes. The data that you wrote to the volumes is not restored. To restore the data on the volumes, you must restore application data from any application that uses the volumes on the clustered system as storage separately. Therefore, you must have a backup of this data before you follow the configuration recovery process.

If the system uses encryption and uses USB flash drives to manage encryption keys, then at least 3 USB flash drives need to be installed in the node canister USB ports for the configuration restore to work. The 3 USB flash drives must be inserted into the node canister or enclosure from which the configuration restore commands are run. Any USB flash drives in other nodes or enclosures (that might become part of the system) are ignored. On systems with fewer than 3 USB ports, encryption must be enabled manually later on in the recovery. On these systems, follow the instructions that are displayed on screen to manually enable encryption during step 14 when the configuration restore is prepared. If you are not recovering an encrypted transparent cloud tiering configuration, the USB flash drives do not need to contain any keys. They are for generation of new keys as part of the restore process. If you are recovering an encrypted transparent cloud tiering configuration, the USB flash drives must contain the previous set of keys to allow the current encrypted data to be unlocked and re-encrypted with the new keys.

During recovery, a new system is created with a new certificate. If the system uses key servers to manage encryption keys, the new system certificate must be exported by using the chsystemcert -export command, then installed on all key servers before the configuration restore operation prepares successfully. It might also be necessary to get the new system's certificate signed if the previous system was using a signed certificate.

About this task

You must regularly back up your configuration data and your application data to avoid data loss. If a system is lost after a severe failure occurs, both configuration for the system and application data is lost. You must restore the system to the exact state it was in before the failure, and then recover the application data.

During the restore process, the nodes and the storage enclosure are restored to the system, and then the MDisks and the array are re-created and configured. If multiple storage enclosures are involved, the arrays and MDisks are restored on the proper enclosures based on the enclosure IDs.

Important:
  • There are two phases during the restore process: prepare and execute. You must not change the fabric or system between these two phases.
  • For systems that contain nodes that are attached to external controllers virtualized by iSCSI, all nodes must be added into the system before you restore your data. Additionally, the system mkip settings and iSCSI storage ports must be manually reapplied before you restore your data.
  • For VMware vSphere Virtual Volumes (sometimes referred to as VVols) environments, after a T4 restoration, some of the Virtual Volumes configuration steps are already completed: metadatavdisk created, usergroup and user created, adminlun hosts created. However, the user must then complete the last two configuration steps manually (creating a storage container on IBM Spectrum® Connect and creating virtual machines on VMware vCenter). See Configuring Virtual Volumes.
  • Restoring the system configuration should be performed via one of the nodes previously in IO group zero. For example, property name="IO_group_id" value="0". The remaining enclosures should be added, as required, in the appropriate order based on the previous IO_group_id of its node canisters.
  • If the system uses encryption, the restore procedure generates new encryption keys. Make sure key servers are online and USB flash drives are installed in the system. Any existing encryption key files on USB flash drives correspond to the previous system and are no longer required, unless using transparent cloud tiering.
  • If the system uses encryption with transparent cloud tiering, the USB flash drive and key servers containing the encryption key for the previous system must be installed in the system before recovery. If the original encryption key is not available, and the system data is encrypted in the cloud provider, then the data in the cloud is not accessible. If the system contains an encrypted cloud account that is configured with both USB and key server encryption, the encryption keys from both need to be available at the time of a T4 recovery.
  • If the system contains an encrypted cloud account that uses USB encryption, a USB flash drive with the system master key must be present in the configuration node before the cloud account can move to the online state. This requirement is necessary when the system is powered down, and then restarted.
  • After a T4 recovery, cloud accounts are in an offline state. It is necessary to re-enter the authentication information to bring the accounts back online.
  • If you use USB flash drives to manage encryption keys, the T4 recovery causes the connection to a cloud service provider to go offline if the USB flash drive is not inserted into the system. To fix this issue, insert the USB flash drive with the current keys into the system.
  • If you use key servers to manage encryption keys, the T4 recovery causes the connection to a cloud service provider to go offline if the key server is offline. To fix this issue, ensure that the key server is online and available during T4 recovery.
  • If you use both key servers and USB flash drives to manage encryption keys, the T4 recovery causes the connection to a cloud service provider to go offline if the key server is offline. To fix this issue, ensure that both the key server is online and a USB flash drive is inserted into the system during T4 recovery.
  • After a T4 recovery, volumes with cloud snapshots that were enabled before the recovery need to have the cloud snapshots manually reenabled.
  • If Fibre Channel port masks are required on the system, these port masks need to be manually reconfigured after a T4 recovery.
  • FlashCopy® mappings and host mappings that have inconsistent ownership groups are not restored.
  • For systems that are previously configured with 128 KiB strip size array configurations, the T4 procedure can still re-create arrays with a 128 KiB strip size even though only a 256 KiB strip size is supported.

If you do not understand the instructions to run the CLI commands, see the command-line interface reference information.

To restore your configuration data, follow these steps:

Procedure

  1. Verify that all nodes are available as candidate nodes before you run this recovery procedure. You must remove errors 550 or 578 to put the node in candidate state. For all nodes that display these errors, follow this procedure.
    1. Point your browser to the service IP address of one of the nodes (for example, https://node_service_ip_address/service/).
    2. Log on to the service assistant.
    3. From the Home page, put the node canister into service state if it is not already in that state.
    4. Select Manage System.
    5. Click Remove System Data.
    6. Confirm that you want to remove the system data when prompted.
    7. Exit service state from the Home page. The 550 or 578 errors are removed, and the node appears as a candidate node.
    8. Remove the system data for the other nodes that display a 550 or a 578 error.

      All nodes previously in this system must have a node status of Candidate and have no errors listed against them.

      Note: A node that is powered off might not show up in this list of nodes for the system. Diagnose hardware problems directly on the node using the service assistant IP address and by physically verifying the LEDs for the hardware components.
      Warning: If you use the management GUI for the initial setup to restore the system configuration, check if a default call home email user was created. If it was created, delete the default call home email user in order for the T4 system recovery to proceed successfully.
  2. Verify that all nodes are available as candidate nodes with blank system fields. Complete the following steps on one node in each control enclosure:
    1. Connect to the service assistant on either of the nodes in the control enclosure.
    2. Select Configure Enclosure.
    3. Select the Reset the system ID option. Do not make any other changes on the panel.
    4. Click Modify.
  3. Create a system by using the technician port.
  4. The setup wizard is shown. Be aware of the following items:
    1. Accept the license agreements.
    2. Set the values for the system name, date and time settings, and the system licensing. The original settings are restored during the configuration restore process.
    3. Verify the hardware. Only the control enclosure on which the clustered system was created and directly attached expansion enclosures are displayed. Any other control enclosures and expansion enclosures in other I/O groups are added to the system later.

    Once the setup wizard finishes, make no other configuration changes.

  5. If you set up email notification in the setup wizard, you must now remove that email user and server so that the original configuration can be restored.
    Issue the following CLI command to remove the new email user:
    rmemailuser 0
    Issue the following CLI command to remove the new email server:
    rmemailserver 0
  6. Optional: From the management GUI, click Access > Users and configure an SSH key for the superuser.
  7. By default, the newly initialized system is created in the storage layer. The layer of the system is not restored automatically from the configuration backup XML file. If the system you are restoring was previously configured in the replication layer, you must change the layer manually now. For more information about the replication layer and storage layer, see the System layers topic in the Related concepts section at the end of the page.
  8. If the clustered system was previously configured as replication layer, then use the chsystem command to change the layer setting.
  9. For configurations with more than one I/O group, add the rest of the control enclosures into the clustered system by using the addcontrolenclosure CLI command.
    The remaining enclosures are added in the appropriate order based on the previous IO_group_id of its node canisters. The following example shows the command to add a control enclosure to I/O group 2.
    svctask addcontrolenclosure -sernum SVT5M48 -iogrp 2
  10. Identify the configuration backup file that you want to restore.

    The file can be either a local copy of the configuration backup XML file that you saved when you backed-up the configuration or an up-to-date file on one of the nodes.

    Configuration data is automatically backed up daily at 01:00 system time on the configuration node.

    Download and check the configuration backup files on all nodes that were previously in the system to identify the one containing the most recent complete backup.

    1. From the management GUI, click Settings > Support > Support Package.
    2. Expand Manual Upload Instructions and select Download Support Package.
    3. On the Download New Support Package or Log File page, select Download Existing Package.
    4. For each node (canister) in the system, complete the following steps:
      1. Select the node to operate on from the selection box at the top of the table.
      2. Find all the files with names that match the pattern svc.config.*.xml*.
      3. Select the files and click Download to download them to your computer.
    The XML files contain a date and time that can be used to identify the most recent backup. After you identify the backup XML file that is to be used when you restore the system, rename the file to svc.config.backup.xml.
  11. Copy onto the system the XML backup file from which you want to restore.
    pscp full_path_to_identified_svc.config.file 
    superuser@cluster_ip:/tmp/svc.config.backup.xml
  12. If the system contains any iSCSI storage controllers, these controllers must be detected manually now. The nodes that are connected to these controllers, the iSCSI port IP addresses, and the iSCSI storage ports must be added to the system before you restore your data.
    Note: If the system contains only Fibre Channel storage controllers, proceed to the next step.
    Note: For a stretched or HyperSwap® topology, after you run the addnode command, change the sites of all of the nodes added in the system. For example,
    chnodecanister -site site_id node_id/node_name
    1. To add these nodes, determine the panel name, node name, and I/O groups of any such nodes from the configuration backup file. To add the nodes to the system, run the following command:
      svctask addcontrolenclosure -iogrp iogrp_name_or_id -sernum enclosure_serial_number -site site_id

      Where enclosure_serial_number is the serial number of the control enclosure, iogrp_name_or_id is the name or ID of the I/O group to which you want to add this node, and site_id is the numeric site value (1 or 2) of the control enclosure.

    2. Run the following command to change the replication layer.
      chsystem -layer replication
    3. To restore iSCSI initiator port IP addresses, use the mkip command. All the IP addresses belonging to the storage portset (i.e portset3) needs to be configured. Find out the IP addresses whose "portset_name" property matches with portset3 in the node_ethernet_ip section from the backup configuration file and restore the IP addresses.
      1. To restore IP address, determine port_ID, node_name, portset_name, IP address, prefix and vlan of the IP address belonging to portset3 from the configuration backup file, and run the following command:
        mkip -node node_id_or_name -port port_id -portset portset_id_or_name -ip ip_address -prefix prefix -vlan vlan

        Where node_name_or_id is the name or id of the node, port_id is the id of the port, portset_id_or_name is the name or id of the portset ip_address is the IP address of the port, prefix is the prefix and vlan is the vlan.

        Complete step i for all (earlier configured) IP ports belonging to portset3 in the node_ethernet_ip section from the backup configuration file.

    4. Next, detect and add the iSCSI storage port candidates by using the detectiscsistorageportcandidate and addiscsistorageport commands. Make sure that you detect the iSCSI storage ports and add these ports in the same order as you see them in the configuration backup file. If you do not follow the correct order, it might result in a T4 failure. Step c.i must be followed by steps c.ii and c.iii. You must repeat these steps for all the iSCSI sessions that are listed in the backup configuration file exactly in the same order.
      1. To detect iSCSI storage ports, determine src_port_id, IO_group_id (optional, not required if the value is 255), target_ipv4/target_ipv6 (the target IP that is not blank is required), iscsi_user_name (not required if blank), iscsi_chap_secret (not required if blank), and site (not required if blank) from the configuration backup file, run the following command:
        svctask detectiscsistorageportcandidate -srcportid src_port_id -iogrp IO_group_id 
        -targetip/targetip6 target_ipv4/target_ipv6 -username iscsi_user_name -chapsecret iscsi_chap_secret -site site_id_or_name

        Where src_port_id is the source Ethernet port ID of the configured port, IO_group_id is the I/O group ID or name being detected, target_ipv4/target_ipv6 is the IPv4/IPv6 target iSCSI controller IPv4/IPv6 address, iscsi_user_name is the target controller user name being detected, iscsi_chap_secret is the target controller chap secret being detected, and site_id_or_name is the specified id or name of the site being detected.

      2. Match the discovered target_iscsiname with the target_iscsiname for this particular session in the backup configuration file by running the lsiscsistorageportcandidate command, and use the matching index to add iSCSI storage ports in step c.iii.

        Run the svcinfo lsiscsistorageportcandidate command and determine the id field of the row whose target_iscsiname matches with the target_iscsiname from the configuration backup file. This is your candidate_id to be used in step c.iii.

      3. To add the iSCSI storage port, determine IO_group_id (optional, not required if the value is 255), site (not required if blank), iscsi_user_name (not required if blank in backup file), and iscsi_chap_secret (not required if blank) from the configuration backup file, provide the target_iscsiname_index matched in step c.ii, and then run the following command:
        addiscsistorageport -iogrp iogrp_id -username iscsi_user_name -chapsecret iscsi_chap_secret -site site_id_or_name candidate_id

        Where iogrp_id is the I/O group ID or name that is added, iscsi_user_name is the target controller user name that is being added, iscsi_chap_secret is the target controller chap secret being added, and site_id_or_name specified the ID or name of the site being that is added.

      4. If the configuration is a HyperSwap or stretched system, the controller name and site needs to be restored. To restore the controller name and site, determine ccontroller_name and controller site_id/name from the backup xml file by matching the inter_WWPN field with the newly added iSCSI controller, and then run the following command:
        chcontroller -name controller_name -site site_id/name controller_id/name

        Where controller_name is the name of the controller from the backup xml file, site_id/name is the ID or name of the site of iSCSI controller from the backup xml file, and controller_id/name is the ID or current name of the controller.

  13. If the system is using Lightweight Directory Access Protocol (LDAP) as the remote authentication service with an administrator password configured, the password must be restored manually before you restore your data. The following example shows the command to configure the LDAP administrator user name and password:
    svctask chldap -username ldap_username -password 'administrator_password'
  14. If the system is using multifactor authentication with IBM Security Verify, the OpenID Connect and API credentials must be restored manually before you restore your data. The following example shows the command to configure the parameters Open ID Client ID, Open ID Client Secret, API Client ID and API Client Secret:
    svctask chauthmultifactorverify -openidclientid 'clientid' -openidclientsecret 'clientsecret' -cliclientid 'clientid' -cliclientsecret 'clientsecret'
  15. If the system is using single sign-on, the OpenID Connect credentials must be restored manually before you restore your data. The following example shows the command to configure the parameters Client ID and Client Secret:
    svctask chauthsinglesignon -clientid 'clientid' -clientsecret 'clientsecret'
  16. If the system has key server encryption, the new certificate must be exported by using the chsystemcert -export command, and then installed on all key servers in the correct device group before you run the T4 recovery. The device group that is used is the one in which the previous system was defined. It might also be necessary to get the new system's certificate signed.
  17. Issue the following CLI command to compare the current configuration with the backup configuration data file:
    svcconfig restore -prepare
    This CLI command creates a log file in the /tmp directory of the configuration node. The name of the log file is svc.config.restore.prepare.log.
    Note: It can take up to a minute for each 256-MDisk batch to be discovered. If you receive error message CMMVC6200W for an MDisk after you enter this command, all the managed disks (MDisks) might not be discovered yet. Allow a suitable time to elapse and try the svcconfig restore -prepare command again.
  18. Issue the following command to copy the log file to another server that is accessible to the system:
    pscp superuser@cluster_ip:/tmp/svc.config.restore.prepare.log 
    full_path_for_where_to_copy_log_files
  19. Open the log file from the server where the copy is now stored.
  20. Check the log file for errors.
    • If you find errors, correct the condition that caused the errors and reissue the command. You must correct all errors before you can proceed to step 21.
    • If you need assistance, contact the support center.
  21. Issue the following CLI command to restore the configuration:
    svcconfig restore -execute

    This CLI command creates a log file in the /tmp directory of the configuration node. The name of the log file is svc.config.restore.execute.log.

  22. Issue the following command to copy the log file to another server that is accessible to the system:
    pscp superuser@cluster_ip:/tmp/svc.config.restore.execute.log 
    full_path_for_where_to_copy_log_files
  23. Open the log file from the server where the copy is now stored.
  24. Check the log file to ensure that no errors or warnings occurred.
    Note: You might receive a warning that states that a licensed feature is not enabled. This message means that after the recovery process, the current license settings do not match the previous license settings. The recovery process continues normally and you can enter the correct license settings in the management GUI later.

    When you log in to the CLI again over SSH, you see this output:

    IBM_2076:your_cluster_name:superuser>
  25. After the configuration is restored, verify that the quorum disks are restored to the MDisks that you want by using the lsquorum command. To restore the quorum disks to the correct MDisks, issue the appropriate chquorum CLI commands.
    Note: If IP Quorum was enabled on the system, it is not recovered automatically as the system certificate is regenerated. It is necessary to manually re-enable IP Quorum by downloading a java application from the Settings>System>IP Quorum tab in the GUI, and then installing the application on the host server.

What to do next

You can remove any unwanted configuration backup and restore files from the /tmp directory on your configuration by issuing the following CLI command:
svcconfig clear -all