© 2017 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public.
Page 136 of 146
SAP HANA Failover Test (Scale-Out Only)
It is very important to test the node failover before you release the system to the customer.
If a SAP HANA node is reset by the master name server, a log entry is written in the /var/log/messages file of the
master name server.
Aug 26 17:24:54 server01 logger:
STONITH HANA Node:server05
Aug 26 17:24:55 server01 logger: anaadm Resetting the HANA Node server05
because of an Nameserver reset command
Aug 26 17:25:58 server01 logger: anaadm HANA Node server05 switched from ON to
OFF
Aug 26 17:27:00 server01 logger: anaadm HANA Node server05 switched from OFF
to ON
Aug 26 17:27:01 server01 logger: anaadm Release NFS locks of HANA Node
server05 done, system is booting
Make sure that all nodes respond to the
ipmitool
command.
Make sure that all management IP addresses are in the /etc/host file of the appliance.
mgmtsrv01:/opt/Cisco/scripts #
grep ipmi /etc/hosts
192.168.76.28 server03-ipmi server03-ipmi
192.168.76.29 server01-ipmi server01-ipmi
192.168.76.37 server04-ipmi server04-ipmi
192.168.76.38 server02-ipmi server02-ipmi
mgmtsrv01:/opt/Cisco/scripts #
If the SAP HANA master name server is “killing” one of the nodes, a status message is written in the
/var/log/messages file of the master name server. The output looks like the following text. The default response of
the STONITH process is to reboot the node.
cishana01:/hana/shared # grep logger /var/log/messages
Aug 17 02:52:13 cishana01 logger:
STONITH HANA Node:cishana02
Aug 17 02:52:13 cishana01 logger: anaadm Resetting the HANA Node
cishana02
because of an Nameserver reset command
Aug 17 02:52:27 cishana01 logger: anaadm HANA Node cishana02 switched from
ON
to OFF
Aug 17 02:52:35 cishana01 logger: anaadm HANA Node cishana02 switched from
OFF
to ON
Aug 17 02:52:35 cishana01 logger: anaadm Release NFS locks of HANA Node
cishana02 done, system is booting
cishana01:/hana/shared #
