Upgrade a Multinode Environment with the Delta Bundle

Note

  • While system upgrade takes approximately two hours at a single site, this may vary in accordance with your topology, number of devices and subscribers. Adjust your upgrade maintenance window to allow for your configuration.

From VOSS-4-UC 18.1 or CUCDM 11.5.3 onwards, the standard screen command should be used where indicated, and the reconnect parameter is available if needed:

  • screen - start a new session
  • screen -ls - show sessions already available
  • screen -r [screen PID] - reconnect to a disconnected session

We recommend using the screen command to avoid failures if the connection is interrupted whilst running the command. If the connection is interrupted whilst running the command in screen then the session can be retrieved by first listing the sessions PID currently running in screen: screen -ls, and then reconnecting to the session using screen -r [screen PID].

The version of screen used in VOSS-4-UC also supports the creation of a log file. If long-running commands will be run, the log file captures screen console output up to the session timeout. A message shows:

timed out waiting for input: auto-logout

To create a screen log file:

  1. Run screen and wait for screen to open.
  2. Press <Ctrl>-a then : (colon). This will enter screen command mode at the bottom of the console.
  3. Create your screen logfile in the media/ directory:
    1. In screen command mode, type logfile media/<screen-logfilename>.log
    2. Press <Enter>
    3. Press <Ctrl>-a and then H to start writing to the log file
    4. Run your commands.

If the screen session times out, you can obtain console output from the log file, for example:

$ sftp platform@<host>:media/<screen-logfilename>.log

Download Files and Check

Description and Steps Notes and Status

VOSS SFTP server: secure.voss-solutions.com

Download XXX-Delta-Bundle.script file from the VOSS SFTP server. Transfer the XXX-Delta-Bundle.script file to the media/ folder of the primary Unified node. Two file transfer options:

Either using SFTP:

  • sftp platform@<primary_unified_node_hostname>
  • cd media
  • put <XXX-Delta-Bundle.script>

Or using SCP:

  • scp <XXX-Delta-Bundle.script> platform@<primary_unified_node_hostname>:~/media

On the primary Unified node, verify that the .script file copied:

  • ls -l media/

On the primary Unified node, verify that the original .sha256 checksums on the SFTP server match.

  • system checksum media/<XXX-Delta-Bundle.script>

    Checksum: <SHA256>

Adaptations Check

Description and Steps Notes and Status

Identify installed adaptations and determine any effect on the upgrade plan.

If the release is accompanied by Upgrade Notes, refer to the details.

Run template customization audits at the sys and sys.hcs hierarchy levels to identify template definitions and instances that were not delivered in the standard template packages during an installation or upgrade.

The audit report includes custom model schema definitions as well as data, domain, and view instances created on the hierarchy node as a result of workflow execution. If the release is accompanied by Upgrade Notes, refer to the details.

  1. Log in as an administrator above Provider level that has access to the hierarchies.
  2. Choose Administration Tools > Reports > Audit Template Customization.
  3. Choose the hierarchy node for which you want to audit customized templates.
  4. Click Save.

View the audit report:

  1. Choose Administration Tools > Reports > Template Customization Reports. A list of template customization audit reports is displayed.
  2. Click a report to view the details. The message field shows how many customized templates were found at the hierarchy node. The details fields lists the model type and instance of each customized template.

Schedules, Transactions and Version Check

Description and Steps Notes and Status

Turn off any scheduled imports to prevent syncs triggering part way through the upgrade. Two options are available:

Individually for each job:

  1. Log in on the GUI as a high level administrator above Provider level.
  2. Select the Scheduling menu to view scheduled jobs.
  3. Click each scheduled job. On the Base tab, uncheck the Activate check box.

Mass modify:

  1. On the GUI, export scheduled syncs into a bulk load sheet.
  2. Modify the schedule settings to de-activate scheduled syncs.
  3. Import the sheet.

Check for running imports. Either wait for them to complete or cancel them:

  1. Log in on the GUI as a high level administrator above Provider level.
  2. Select the Transaction menu to view transactions.
  3. Filter the Action column:
    1. Choose Status as “Processing” and then choose each Action that starts with “Import”, for example, “Import Unity Connection”.
    2. Click Search and confirm there are no results.
    3. If there are transactions to cancel, select them and click Cancel.

Record the current version information. This is required for upgrade troubleshooting.

  • Log in on the GUI and record the information contained in the About > Extended Version

Pre-Upgrade, Security and Health Steps

Description and Steps Notes and Status

Verify that the primary node is the active primary node at the time of upgrade.

database config

Ensure that the node on which the installation will be initiated has the stateStr parameter set to PRIMARY and has the highest priority number (highest priority number could vary depending on cluster layout).

Example output

<ip address>:27020:
  priority: <number>
  stateStr: PRIMARY
  storageEngine: WiredTiger

Validate the system health.

On the Primary Unified Node, verify cluster connectivity:

  • cluster status

On each node verify network connectivity, disk status and NTP.

  • cluster check

If there is any sign of the paths below are over 80% full, a clean-up is needed to avoid risk of for example full logs occurring during upgrade. Clean-up steps are indicated next to the paths:

/              (call support if over 80%)
/var/log       (run: log purge)
/opt/platform  (remove any unnecessary files from /media directory)
/tmp           (reboot)

On the Primary Unified Node, verify there are no pending Security Updates on any of the nodes:

  • cluster run all security check

Shutdown servers and take snapshots from VMWare and then power on all servers, starting with the primary:

Use VMware snapshots. Consider the following:

  • VOSS cannot guarantee that a VMware snapshot can be used to successfully restore VOSS-4-UC or any Cisco HCS Management application. If you cannot restore the application from a snapshot, your only recourse is to reinstall the application.
  • When the backup is complete and you do not need the VMware snapshot for restore activities, delete the snapshot immediately to preserve LUN space.
  • cluster run notme system shutdown –force && system shutdown –force

Log into VMWare and take snapshots of all unified nodes and all web proxies.

After snapshots, restart the servers:

  • Power up the servers via VMWare.

Optional: If a backup is required in addition to the snapshot, use the backup add <location-name> and backup create <location-name> commands. For details, refer to the Platform Guide.
Description and Steps Notes and Status

Before upgrading, check all services, nodes and weights for the cluster:

Make sure no services are stopped/broken. The message ‘suspended waiting for mongo’ is normal on the fresh unified nodes.

  • cluster run all app status

Make sure all application nodes show 3 or 5 nodes.

  • cluster run application cluster list

Check that the database weights are set. It is critical to ensure the weights are set before upgrading a cluster.

  • cluster run application database weight list

Example output:

172.29.21.240:
    weight: 80
172.29.21.241:
    weight: 70
172.29.21.243:
    weight: 60
172.29.21.244:
    weight: 50

Verify the primary node in the primary site and ensure no nodes are in the ‘recovering’ state (stateStr is not RECOVERING). On the primary node:

  • database config

Upgrade

Description and Steps Notes and Status

On the primary unified node:

  • screen

Run (optionally with command parameters below):

  • app install media/<script_file> delete-on-success yes –force

From release 19.1.2 and later, the delete-on-success parameter and yes or no value have been added to remove or keep the the script file in the media/ directory after successful installation.

Post-Upgrade, Security and Health Steps

Description and Steps Notes and Status

On the primary unified node, verify the cluster status:

  • cluster status

  • cluster check

  • If any of the above commands show errors, check for further details to assist with troubleshooting:

    cluster run all diag health
If upgrade is successful, the screen session can be closed by typing exit in the screen terminal. If errors occurred, keep the screen terminal open for troubleshooting purposes and contact VOSS support.

Check for needed security updates. On the primary node, run:

  • cluster run all security check

If one or more updates are required for any node, run on the primary Unified node:

  • cluster run all security update

Note: if the system reboots, do not carry out the next manual reboot step.

Manual reboot only if needed:

  • cluster run notme system reboot

If node messages: <node name> failed with timeout are displayed, these can be ignored.

  • system reboot

Since all services will be stopped, this takes some time.

Database Filesystem Conversion

Important

This step is to be carried out only if you have not converted the file system before.

To check if the step is not required:

  1. Run database config and ensure that the storage engine for all database nodes shows as storageEngine: WiredTiger.
  2. Run drives list and ensure that the LVM storage shows for all converted database nodes under Volume Groups.

The database convert_drive command provides parameters that allow for a flexible upgrade schedule in order to limit system downtime.

When the database convert_drive command is run, the voss-deviceapi service will be stopped first and started after completion. The command should therefore be run during a maintenance window while there are no running transactions.

The procedure and commands in this step depend on:

  • your topology
  • latency between data centers
  • upgrade maintenance windows - Window 1 to Window 3 represent chosen maintenance windows.

For the Database Filesystem Conversion step below, first inspect the table below for guidance on the commands to run according to your configuration and preferences.

  • Run all commands on the primary unified node:

    • Ensure states of database nodes are not DOWN - otherwise the command will fail:

      database config (stateStr is not DOWN)

    • Ensure database weights are set and there is 1 maximum weight - otherwise the command will fail:

      database weight list (one weight value is maximum)

  • For 2 and 3 maintenance windows: after the upgrade (prior to Windows 2 and 3), only nodes with converted drives will generate valid backups.

    For example, if the primary drive is converted, backups from the primary node can be used to restore the database. If there is a database failover to the highest weight secondary node that was not converted, it will not be possible for backups to be generated on that secondary node until the drive is converted.

Note

The database convert_drive command can also be run on a single node only by running the following command and parameter from the specific node: database convert_drive standalone. This option can for example be used for performance reasons in cases where a node is in a remote location.

Topology Window 1 Window 2 Window 3 Commands (DC = valid data center name) Description
multinode Y    

database convert_drive secondary all

database convert_drive primary

 Recommended for a system with latency < 10ms.
multinode Y Y  

Window 1:

database convert_drive primary

Window 2:

database convert_drive secondary all

Can be used for a system with latency < or > 10ms.

Allows for smaller maintenance windows.

Cluster is not available during maintenance.
multinode Y Y Y

Window 1:

database convert_drive primary

Window 2:

database convert_drive secondary <first DC>

Window 3:

database convert_drive secondary <second DC>

Can be used for a system with latency > 10ms.

Allows for smaller maintenance windows.

Cluster is not available during maintenance.
Description and Steps Notes and Status

Database Filesystem Conversion step

Shut down all the nodes. Since all services will be stopped, this takes some time.

  • cluster run notme system shutdown –force && system shutdown –force

Create a VMWare snapshot for all the unified servers so that the system can easily be reverted in the case of a conversion error. Boot all the systems in VMWare.

  • Run the convert_drive command with parameters according to the table above.

    Wait until it completes successfully.

  • database config

    Ensure that the storage engine for all converted database nodes shows as storageEngine: WiredTiger

  • drives list

    Ensure that the LVM storage shows for all converted database nodes under Volume Groups

In the example below, dbroot/dm-0 shows under Volume Groups, Logical volumes

$ drives list
Used disks and mountpoints:
sdc1 - services:backups
dm-0 - mongodb:dbroot

Unused disks:
none - if disks have been hot-mounted, it may be necessary to reboot the system

Unused mountpoints:
services:SWAPSPACE

Volume Groups
voss - 10.0 GB free, 60.0 GB total
Physical volumes:
sdd1
Logical volumes:
dbroot/dm-0 - 50.0 GB

Post Template Upgrade Tasks

Description and Steps Notes and Status

Verify the upgrade:

Log in on the GUI and check the information contained in the About > Extended Version menu. Confirm that versions have upgraded:

  • Release should show 19.3.2
  • Platform Version should show 19.3.2

If your web browser cannot open the user interface, clear your browser cache before trying to open the interface again.

Restore Adaptations

Description and Steps Notes and Status

Restore and adaptations prior to upgrade.

If the release is accompanied by Upgrade Notes, refer to the details on adaptation impact.

Restore Schedules

Description and Steps Notes and Status

Re-enable scheduled imports if any were disabled prior to the upgrade. Two options are available:

Individually for each job:

  1. Log in on the GUI as a high level administrator above Provider level.
  2. Select the Scheduling menu to view scheduled jobs.
  3. Click each scheduled job. On the Base tab, check the Activate check box.

Mass modify:

  1. Modify the exported sheet of schedules to activate scheduled syncs.
  2. Import the bulk load sheet.

Log Files and Error Checks

Description and Steps Notes and Status

Inspect the output of the command line interface for upgrade errors.

Use the log view command to view any log files indicated in the error messages, for example, run the command if the following message appears:

For more information refer to the execution log file with
'log view platform/execute.log'

For example, if it is required send all the install log files in the install directory to an SFTP server:

  • log send sftp://x.x.x.x install
Log in on the GUI as system level administrator, go to Administration Tools > Transaction and inspect the transactions list for errors.