Skip to content

Rolling restart of services

To reduce or eliminate downtime for services with multiple instances, Control Center restarts instances of the service one at a time.

The Control Center browser interface visually indicates when a service is restarting and whether an instance is down during the restart. When a wide area network (WAN) outage occurs, the rolling restart proceeds, with instances on the disconnected hosts restarting when the WAN is restored.

The serial process applies to restart only. When stopping and starting services, Control Center starts and stops all instances immediately.

Optionally, you can specify that Control Center is to stop all instances of a service before restarting.

Changing rolling restart

Use this procedure to stop all instances of a service before restarting, instead of performing a rolling restart.

  1. Log in to the Control Center master host as root or as a user with superuser privileges.
  2. Restart all instances of the service with the --rebalance option.

    Replace SERVICE with the name of the service to restart:

    for service in $(serviced service list --show-fields ServiceID,Name | awk '/SERVICE/ {print $1}')
      serviced service restart --rebalance $service