Pepperdata pepper logo
Support Logout

Rollback a Pepperdata Upgrade

The Pepperdata rollback procedure is the same as the upgrade procedure, except that instead of replacing Pepperdata with a later version, you replace it with an earlier version. The process is quicker if you did not remove the previous Pepperdata package during the upgrade. Instead of manually stopping the Pepperdata agents, removing the old Pepperdata software, and installing the new Pepperdata software, you can run one of the Pepperdata version configuration scripts.

The version configuration scripts change the version of Pepperdata that Hadoop uses from the latest installed Pepperdata software to any given version that is installed on a given host. The scripts use Linux alternatives to reconfigure the active Pepperdata version.

On This Page

Run all commands as the root user.

Task 1: Configure Hadoop to Run the Old Pepperdata Version

Run the applicable command from the table, based on which version of Pepperdata you want to run.

Result Alternatives Command
Force version X.Y.Z to be the active version of the Pepperdata Supervisor (including PepAgents and monitored services). /opt/pepperdata/supervisor-X.Y.Z/alternatives/set-current-agents
Force the currently active version to remain the active version of the Pepperdata Supervisor (including PepAgents and monitored services). /opt/pepperdata/supervisor/alternatives/set-current-agents
Revert to default behavior where the newest version of the Pepperdata Supervisor agents are active. /opt/pepperdata/supervisor-X.Y.Z/alternatives/use-latest-agents
Force version X.Y.Z to be the active version of the PepCollector. /opt/pepperdata/collector-X.Y.Z/alternatives/set-current-collectd
Force the current active version to remain the active version of the PepCollector. /opt/pepperdata/collector/alternatives/set-current-collectd
Revert to default behavior where the newest version of the PepCollector is active. /opt/pepperdata/collector-X.Y.Z/alternatives/use-latest-collectd

You can have a newer active PepCollector version than the active Supervisor version, but not the opposite. To run with different PepCollector and Supervisor versions, contact Pepperdata Support.

Task 2: Restart Hadoop YARN Services

Procedure

  1. Restart the affected YARN services:

    • YARN ResourceManager
    • YARN NodeManagers

  2. (If using HBase) Restart the HBase RegionServers.

  3. (If using Query Spotlight) On every Hive server, restart the hiveserver2 service.

Task 3: Restart the Pepperdata Agents

Procedure

  1. Start the Pepperdata Collector.

    You can use either the service (if provided by your OS) or systemctl command:

    • sudo service pepcollectd start
    • sudo systemctl start pepcollectd

    If any of the process’s startup checks fail, an explanatory message appears and the process does not start. Address the issues and try again to start the process.

  2. Start the PepAgent.

    You can use either the service (if provided by your OS) or systemctl command:

    • sudo service pepagentd start
    • sudo systemctl start pepagentd

    If the service successfully starts, the Pepperdata Version [VERSION-STRING] OK message appears in the /var/log/pepperdata/pepagent/pepagent.log file.

    If any of the process’s startup checks fail, an explanatory message appears and the process does not start. Address the issues and try again to start the process.