PowerFlex: How to replace a node member in the MDM Cluster
Summary: Replace a member of an MDM cluster to replace a faulty server or change the server IP address.
Instructions
If you have an extra server to replace the current server, then there is no need to change the cluster mode (3-node or 5-node). If you do not have an additional server, you must reduce the cluster mode, from 5-node to 3-node, or from 3-node to single node.
- To remove a cluster member, you first must make it a standby, then remove the standby. To add a member to a cluster, you first make it a standby, then add the standby to the cluster.
- The cluster must always have 5, 3, or 1 members, never any other amount. For a further understanding of this subject, see "The MDM cluster" in the Architecture section of Getting to Know Dell PowerFlex Guide. Go to the section that describes your environment:
- "Replace a cluster member by adding a new server"
- "Replace a cluster member without adding a new server to the cluster"
Add or replace a server to the MDM cluster.
Before you begin, perform the following: Assign the necessary IP addresses to the replacement server.
Install the MDM package on the server.
In this example, we are replacing the server whose IP address is 10.3.1.179, currently a member of a 5-node MDM cluster, with a server whose IP address is 10.3.1.57, which is currently external to any PowerFlex system. This process can be used to replace any role in the MDM cluster.
Steps
1- Log in to the Primary MDM with admin login. use single quotes ' ' if password has special characters
3.x
scli --login --username admin --password '<password>'
4.x
scli --login --username admin --password '<password>' --management_system_ip <pfmp_ip>
2- Check that the current server (179) is not the primary MDM by running the following command:
scli --query_cluster
The output similar to the following is displayed:
# scli --query_cluster
Cluster:
Mode: 5_node, State: Normal, Active: 5/5, Replicas: 3/3
Master MDM:
Name: mdm17, ID: 0x5d07497754427fd0
IPs: 10.3.1.17, 192.168.1.17, Management IPs: 10.3.1.17, Port: 9011
Version: 2.0.972
Slave MDMs:
Name: mdm19, ID: 0x26ee566356362451
IPs: 10.3.1.19, 192.168.1.19, Management IPs: 10.3.1.19, Port: 9011
Status: Normal, Version: 2.0.972
Name: mdm18, ID: 0x5843c4d16d8f1082
IPs: 10.3.1.18, 192.168.1.18, Management IPs: 10.3.1.18, Port: 9011
Status: Normal, Version: 2.0.972
Tie-Breakers:
Name: mdm179, ID: 0x7380b70e2f73d346
IPs: 10.3.1.179, 192.168.1.179, Port: 9011
Status: Normal, Version: 2.0.972
Name: mdm20, ID: 0x6dfe1c5f4062b5b3
IPs: 192.168.1.20, 10.3.1.20, Port: 9011
Status: Normal, Version: 2.0.972
In this case, Node name mdm179 is a tiebreaker.
3- If the current server is the primary MDM, change its state using the switch_mdm_ownership command
For PowerFlex 3.5 See Dell EMC PowerFlex v3.5.x CLI Reference Guide
For PowerFlex 4.X See Dell PowerFlex 4.5.x CLI Reference Guide
4- Make the replacement MDM server a standby MDM, and assign it a name (mdm57, in our example) by running the following command, on the primary MDM:
scli --add_standby_mdm --mdm_role tb --new_mdm_ip 10.3.1.57,192.168.1.57 --new_mdm_management_ip 10.3.1.57 --new_mdm_name mdm57
5- You can see the result of the command by running the following command:
scli --query_cluster
The output similar to the following is displayed:
# scli --query_cluster
Cluster:
Mode: 5_node, State: Normal, Active: 5/5, Replicas: 3/3
...
Tie-Breakers:
Name: mdm179, ID: 0x7380b70e2f73d346
IPs: 10.3.1.179, 192.168.1.179, Port: 9011
Status: Normal, Version: 2.0.972
Name: mdm20, ID: 0x6dfe1c5f4062b5b3
IPs: 192.168.1.20, 10.3.1.20, Port: 9011
Status: Normal, Version: 2.0.972
Standby MDMs:
Name: mdm57, ID: 0x073e4c8b1d20d124, Tie Breaker
IPs: 10.3.1.57, 192.168.1.57, Port: 9011
mdm57 has been added as a standby MDM. When it is a standby MDM, it can be added to the cluster.
6- Replace the current mdm179 with the standby mdm57 by running the following command:
scli --replace_cluster_mdm --remove_tb_name mdm179 --add_tb_name mdm57
The following output is displayed:
Successfully replaced the cluster MDM
The current server has been replaced.
Additional Information
Configure and Customize Dell PowerFlex v3.6.x
In PowerFlex 3.6 we use the terms Master MDM and Slave MDM
#scli --switch_cluster_mode --cluster_mode 3_node --remove_slave_mdm_name --remove_tb_name -- #scli --switch_cluster_mode --cluster_mode 5_node --add_slave_mdm_name --add_tb_name
In PowerFlex 4.X we use the terms Primary MDM and Secondary MDM
#scli --switch_cluster_mode --cluster_mode 3_node --remove_secondary_mdm_name --remove_tb_name #scli --switch_cluster_mode --cluster_mode 5_node --add_secondary_mdm_name --add_tb_name