IBM Support

Safely rebooting host servers - PureData System for Analytics

Question & Answer


Question

How do I safely reboot both hosts in a PureData System for Analytics environment?

Answer

You can use the following procedure to safely reboot both PureData System for Analytics N1000/1001, N2001/2002 and N3001-002 or larger appliance hosts and test migration at the same time.

Note: You must have root access available as well as nz user access.

This article assumes that HA1 is the active host and HA2 is the standby host. Verify active and standby hosts with the crm_resource -r nps -W command: This command must be run as root user.

For example, the following output (resource nps is running on: nz55555-h1) indicates that the active host is 'nz55555-h1':

[root@NZ55555-H1 ~]# crm_resource -r nps -W
crm_resource[16959]: 2011/06/15_14:52:09 info: Invoked: crm_resource -r nps -W
resource nps is running on: nz55555-h1

*Before continuing, verify the health of the cluster with the following 2 commands as root:
crm_mon -1
/opt/nz-hwsupport/install_tools/hb_doctor
Should either of the above commands report an error, stop and engage NZ Support.

Note: Connect to the physical (not virtual) address while executing the following procedure:

1. Stop the database on the active host (HA1) as 'nz' user:

  • [nz@NZ55555-H1 ~]#nzstop

2. Reboot the standby host (HA2) as 'root' user:

  • [root@NZ55555-H2 ~]#service heartbeat stop

    [root@NZ55555-H2 ~]#service drbd stop

    [root@NZ55555-H2 ~]#shutdown -r now



3. Once the standby host has rebooted and is reachable (verify HERE), wait 5 minutes for the cluster services to complete startup and then execute below commands as 'root' user to migrate/relocate the database:

  • [root@NZ55555-H1 ~]#/nzlocal/scripts/heartbeat_admin.sh --migrate

Monitor cluster status with the following command to verify that HA2 has become the active host:

  • [root@NZ55555-H1 ~]#crm_mon -i5

Note: You can run above command to monitor cluster status from either hosts, although on the active host you must run it from different terminal.

4. Stop the database as 'nz' user:
  • [nz@NZ55555-H2 ~]#nzstop

5. Reboot HA1 using similar methods as in step 2 as 'root' user.
It should now be the standby host since the database was migrated to HA2 in step 3 above:


6. After HA1 has rebooted and is back in the cluster as standby (verify health using similar methods as before in step 3), migrate back to HA1:

[root@NZ55555-H2 ~]#/nzlocal/scripts/heartbeat_admin.sh --migrate

7.   Monitor cluster status at five second intervals using the following command:
  • [root@NZ55555-H2 ~]#crm_mon -i5

[{"Product":{"code":"SSULQD","label":"IBM PureData System"},"Business Unit":{"code":"BU053","label":"Cloud & Data Platform"},"Component":null,"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"1.0.0","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]

Historical Number

NZ278564

Document Information

Modified date:
13 February 2020

UID

swg21517071