cancel
Showing results for 
Search instead for 
Did you mean: 
cancel

DCNM 10.x - native HA setup

1976
Views
5
Helpful
4
Comments

Why do you need native HA:

 The native HA feature allows two Cisco DCNM appliances to run as active and standby applications, with their embedded databases synchronized in real time. Therefore, when the active DCNM is not functioning, the standby DCNM will take over with the same database data and resume the operation.

 

Limitations of native HA:

  1. The native HA is only supported on DCNM appliances with ISO or OVA installation. 
  2. Once you enable native HA you could not revert back. The only option to have stand-alone server again is to restart DCNM

Licensing

When you setup native HA, you need to license only one DCNM server (the one which has been deployed as Active). During failover, Standby server will get the licensing information from the original Active and will work as a licensed server.

How to setup native HA

Instructions 

 

Please pay attention, the following values in the ha-setup.properties file should be changed: NODE_ID, PERR_ETH0_IP, VIP_ADDRESS, VIP_PREFIX and VIP_FQDN.  If you use Enhanced Fabric Management network, then you also should specify values for VIP1_ADDRESS and VIP1_PREFIX

 

Split brain and HA-PING

 Split-Brain syndrome occurs when the communication on enhanced fabric interface between two HA peers is lost. As the result, both hosts will act as Active. When the communication resumes, both hosts shall negotiate and eventually one will become active, the other standby

 

In order to avoid the split-brain situations, you could configure optional parameter HA_PING in ha-setup.properties .  The address which you configure as HA_Ping should be a 3-rd machine address, i.e. it should not belong one of the DCNM HA instances. The HA_PING address should be reachable from both DCNM servers. It is recommended to use a default-gateway as HA-Ping address.

Q&A

Question: How I could change the configured network mask (VIP_PREFIX) after ha has been setup?

Answer: please edit /etc/ha.d/haresources and changes virtual network mask there. The file would look like this one:

 [root@dcnm1]# cat /etc/ha.d/haresources

dcnm.cisco.com            IPaddr::10.10.10.3/24/eth0  IPaddr::10.48.76.196/8/eth1    monit-fms

 The /24 is a parameter specified as VIP_PREFIX, /8 is a parameter specified as VIP1_PREFIX.  The change should be done on both servers. Once the value has been changed both servers should be rebooted..

 

Question: How I could change the HA_PING address after ha has beet setup?

Answer:

1) edit the ip address in /usr/local/cisco/dcm/fm/ha/ha-ping.conf

2) chkconfig heartbeat off

3) service heartbeat stop

4) echo "* * * * * root /sbin/ha-ping.sh" > /etc/cron.d/ha-ping

5) sed -i "s/APP_STATUS_HEARTBEAT=.*/APP_STATUS_HEARTBEAT=ha-ping-on/g" /root/.DO_NOT_DELETE

 

Question: How I could disable the HA_PING after ha has been setup?

Answer: HA_PING can be disabled as follows (to be done on both nodes):

1) rm /etc/cron.d/ha-ping

2) sed -i "s/APP_STATUS_HEARTBEAT=.*/APP_STATUS_HEARTBEAT=auto/g" /root/.DO_NOT_DELETE

3) service heartbeat start

4) chkconfig heartbeat on

 

Question: How I could revert back from ha setup to standalone servers?

Answer: going from HA to standalone is not supported and there is no easy way to do so. Reinstalling DCNM from scratch is the only  option at the moment.

Comments
Rick1776
Contributor

This is a great article. I have a customer with the same issue and not knowing how to enable HA. It's interesting that you cannot revert back to a non HA setup. 

 

Thanks again.

Is cluster over WAN supported?

ashpatel
Cisco Employee

Nice!

prasadkumbhar
Beginner

What is the use of this feature? if we ping the Default gateway independently, how other/secondary DCNM come to know primary is alive?  if it works as a


@Svetlana Radzevich wrote:

Why do you need native HA:

 The native HA feature allows two Cisco DCNM appliances to run as active and standby applications, with their embedded databases synchronized in real time. Therefore, when the active DCNM is not functioning, the standby DCNM will take over with the same database data and resume the operation.

 

Limitations of native HA:

  1. The native HA is only supported on DCNM appliances with ISO or OVA installation. 
  2. Once you enable native HA you could not revert back. The only option to have stand-alone server again is to restart DCNM

Licensing

When you setup native HA, you need to license only one DCNM server (the one which has been deployed as Active). During failover, Standby server will get the licensing information from the original Active and will work as a licensed server.

How to setup native HA

Instructions 

 

Please pay attention, the following values in the ha-setup.properties file should be changed: NODE_ID, PERR_ETH0_IP, VIP_ADDRESS, VIP_PREFIX and VIP_FQDN.  If you use Enhanced Fabric Management network, then you also should specify values for VIP1_ADDRESS and VIP1_PREFIX

 

Split brain and HA-PING

 Split-Brain syndrome occurs when the communication on enhanced fabric interface between two HA peers is lost. As the result, both hosts will act as Active. When the communication resumes, both hosts shall negotiate and eventually one will become active, the other standby

 

In order to avoid the split-brain situations, you could configure optional parameter HA_PING in ha-setup.properties .  The address which you configure as HA_Ping should be a 3-rd machine address, i.e. it should not belong one of the DCNM HA instances. The HA_PING address should be reachable from both DCNM servers. It is recommended to use a default-gateway as HA-Ping address.

Q&A

Question: How I could change the configured network mask (VIP_PREFIX) after ha has been setup?

Answer: please edit /etc/ha.d/haresources and changes virtual network mask there. The file would look like this one:

 [root@dcnm1]# cat /etc/ha.d/haresources

dcnm.cisco.com            IPaddr::10.10.10.3/24/eth0  IPaddr::10.48.76.196/8/eth1    monit-fms

 The /24 is a parameter specified as VIP_PREFIX, /8 is a parameter specified as VIP1_PREFIX.  The change should be done on both servers. Once the value has been changed both servers should be rebooted..

 

Question: How I could change the HA_PING address after ha has beet setup?

Answer:

1) edit the ip address in /usr/local/cisco/dcm/fm/ha/ha-ping.conf

2) chkconfig heartbeat off

3) service heartbeat stop

4) echo "* * * * * root /sbin/ha-ping.sh" > /etc/cron.d/ha-ping

5) sed -i "s/APP_STATUS_HEARTBEAT=.*/APP_STATUS_HEARTBEAT=ha-ping-on/g" /root/.DO_NOT_DELETE

 

Question: How I could disable the HA_PING after ha has been setup?

Answer: HA_PING can be disabled as follows (to be done on both nodes):

1) rm /etc/cron.d/ha-ping

2) sed -i "s/APP_STATUS_HEARTBEAT=.*/APP_STATUS_HEARTBEAT=auto/g" /root/.DO_NOT_DELETE

3) service heartbeat start

4) chkconfig heartbeat on

 

Question: How I could revert back from ha setup to standalone servers?

Answer: going from HA to standalone is not supported and there is no easy way to do so. Reinstalling DCNM from scratch is the only  option at the moment.



heartbeat.

Content for Community-Ad