Red Hat Summit Red Hat SummitSuporteConsoleDesenvolvedoresComeçar um testeContato

Este conteúdo não está disponível no idioma selecionado.

Chapter 6. Configuring a quorum device in the cluster

We recommend that you configure a qdevice in your cluster for improved service resiliency. Alternatively you can configure a dedicated cluster node that only serves for adding a quorum vote.

Warning

Do not configure both a qdevice and a majority-maker node in the same cluster. This adds one vote per method and results in an even number of quorum votes again.

6.1. Configuring a qdevice for cluster quorum
Copiar o link

If you prefer to use a dedicated majority-maker cluster node for this purpose, skip the qdevice setup and follow the steps in Configuring a majority-maker node for cluster quorum instead.

6.1.1. Preparing the quorum device host
Copiar o link

  • Configure the RHEL High Availability repository on the quorum device host.
  • If the firewalld service is installed and you are not using it on your hosts, disable the service. See Disabling the firewalld service.

6.1.2. Configuring a qdevice on a quorum device host
Copiar o link

At first you must configure a quorum device host that serves the qdevice for your cluster quorum.

In the following steps the example name of the qdevice host is dc3qdevice.

Prerequisites

  • You have installed a separate host that is ideally located in a different location or availability zone than your cluster nodes.
  • You have configured the RHEL High Availability repository on the dedicated quorum device host.
  • You have configured your network in a way that your cluster nodes can reach the quorum device host.

Procedure

  1. Install pcs and corosync-qnetd on the quorum device host: 

    [root]# dnf install pcs corosync-qnetd

  2. Start and enable the pcsd service on the quorum device host: 

    [root]# systemctl enable --now pcsd.service

  3. Create the qdevice on the quorum device host. This command configures and starts the quorum device model net and configures the device to start on boot. Run this command on the quorum device host: 

    [root]# pcs qdevice setup model net --enable --start
Quorum device 'net' initialized
quorum device enabled
Starting quorum device...
quorum device started

  4. Optional: If you are running the firewalld service, enable the ports that are required by the Red Hat High Availability Add-On. Run this on the quorum device host: 

    [root]# firewall-cmd --add-service=high-availability
[root]# firewall-cmd --runtime-to-permanent

  5. Set a password for the user hacluster on the quorum device host: 

    [root]# passwd hacluster

Verification

  • Check the quorum device status on the quorum device host: 

    [root]# pcs qdevice status net --full
QNetd address:                  *:5403
TLS:                            Supported (client certificate required)
Connected clients:              0
Connected clusters:             0
Maximum send/receive size:      32768/32768 bytes

6.1.3. Configuring a qdevice in the cluster
Copiar o link

Prerequisites

  • You have configured a quorum device host that is ideally located in a different location or availability zone than your cluster nodes, for example, dc3qdevice.
  • You have configured a qdevice on the quorum device host.
  • You have configured your network in a way that your cluster nodes can reach the quorum device host.

Procedure

  1. Add the qdevice host to the /etc/hosts on all existing cluster nodes, so that the resulting /etc/hosts entries are the same on all nodes. This ensures that the cluster nodes can communicate with the host even when the DNS services failed: 

    [root]# cat /etc/hosts
...
192.168.100.101 dc1hana1.example.com dc1hana1
192.168.100.102 dc1hana2.example.com dc1hana2
192.168.100.103 dc2hana1.example.com dc2hana1
192.168.100.104 dc2hana2.example.com dc2hana2
192.168.100.120 dc3qdevice.example.com dc3qdevice

  2. Install corosync-qdevice on all nodes of your cluster: 

    [root]# dnf install corosync-qdevice

  3. Authenticate the quorum device host in the cluster to enable communication. Run this command on one cluster node: 

    [root]# pcs host auth <qdevice_host>
Username: hacluster
Password:
dc3qdevice: Authorized
    • Replace <qdevice_host> with the name of your quorum device host, for example, dc3qdevice.

  4. Add the qdevice from the quorum device host to the cluster. Run this command on one cluster node: 

    [root]# pcs quorum device add model net host=<qdevice_host> algorithm=ffsplit
Setting up qdevice certificates on nodes...
dc2hana1: Succeeded
dc1hana2: Succeeded
dc2hana2: Succeeded
dc1hana1: Succeeded
Enabling corosync-qdevice...
dc2hana1: corosync-qdevice enabled
dc1hana1: corosync-qdevice enabled
dc1hana2: corosync-qdevice enabled
dc2hana2: corosync-qdevice enabled
Sending updated corosync.conf to nodes...
dc2hana1: Succeeded
dc1hana1: Succeeded
dc1hana2: Succeeded
dc2hana2: Succeeded
dc1hana1: Corosync configuration reloaded
Starting corosync-qdevice...
dc2hana1: corosync-qdevice started
dc1hana2: corosync-qdevice started
dc1hana1: corosync-qdevice started
dc2hana2: corosync-qdevice started
    • Replace <qdevice_host> with the name of your quorum device host, for example, dc3qdevice.
    • The algorithm can be ffsplit or lms. Consult the corosync-qdevice(8) man page for more details about the different algorithms.

Verification

  1. Check the quorum configuration on a cluster node: 

    [root]# pcs quorum config
Device:
  votes: 1
  Model: net
    algorithm: ffsplit
    host: dc3qdevice

  2. Check the quorum status on a cluster node: 

    [root]# pcs quorum status
Quorum information
------------------
Date:             Wed Sep  3 14:01:24 2025
Quorum provider:  corosync_votequorum
Nodes:            4
Node ID:          1
Ring ID:          1.180
Quorate:          Yes

Votequorum information
----------------------
Expected votes:   5
Highest expected: 5
Total votes:      5
Quorum:           3
Flags:            Quorate Qdevice

Membership information
----------------------
    Nodeid      Votes    Qdevice Name
         1          1    A,V,NMW dc1hana1 (local)
         2          1    A,V,NMW dc1hana2
         3          1    A,V,NMW dc2hana1
         4          1    A,V,NMW dc2hana2
         0          1            Qdevice

  3. Check the quorum device status on a cluster node: 

    [root]# pcs quorum device status
Qdevice information
-------------------
Model:                  Net
Node ID:                1
Configured node list:
    0   Node ID = 1
    1   Node ID = 2
    2   Node ID = 3
    3   Node ID = 4
Membership node list:   1, 2, 3, 4

Qdevice-net information
----------------------
Cluster name:           hana-scaleout-cluster
QNetd host:             dc3qdevice:5403
Algorithm:              Fifty-Fifty split
Tie-breaker:            Node with lowest node ID
State:                  Connected

  4. Check the quorum device status on the quorum device host. The status now shows the details of the cluster on which the qdevice is used. If the same qdevice is configured in multiple clusters, the status contains the details for each cluster: 

    [root]# pcs qdevice status net --full
QNetd address:                  *:5403
TLS:                            Supported (client certificate required)
Connected clients:              4
Connected clusters:             1
Maximum send/receive size:      32768/32768 bytes
Cluster "hana-scaleout-cluster":
    Algorithm:          Fifty-Fifty split (KAP Tie-breaker)
    Tie-breaker:        Node with lowest node ID
    Node ID 1:
        Client address:         ::ffff:10.99.30.121:53312
        HB interval:            8000ms
        Configured node list:   1, 2, 3, 4
        Ring ID:                1.180
        Membership node list:   1, 2, 3, 4
        Heuristics:             Undefined (membership: Undefined, regular: Undefined)
        TLS active:             Yes (client certificate verified)
        Vote:                   ACK (ACK)
    Node ID 2:
        Client address:         ::ffff:10.99.30.57:63610
        HB interval:            8000ms
        Configured node list:   1, 2, 3, 4
        Ring ID:                1.180
        Membership node list:   1, 2, 3, 4
        Heuristics:             Undefined (membership: Undefined, regular: Undefined)
        TLS active:             Yes (client certificate verified)
        Vote:                   ACK (ACK)
    Node ID 3:
        Client address:         ::ffff:10.99.30.120:26758
        HB interval:            8000ms
        Configured node list:   1, 2, 3, 4
        Ring ID:                1.180
        Membership node list:   1, 2, 3, 4
        Heuristics:             Undefined (membership: Undefined, regular: Undefined)
        TLS active:             Yes (client certificate verified)
        Vote:                   ACK (ACK)
    Node ID 4:
        Client address:         ::ffff:10.99.30.158:54932
        HB interval:            8000ms
        Configured node list:   1, 2, 3, 4
        Ring ID:                1.180
        Membership node list:   1, 2, 3, 4
        Heuristics:             Undefined (membership: Undefined, regular: Undefined)
        TLS active:             Yes (client certificate verified)
        Vote:                   No change (ACK)

6.2. Configuring a majority-maker node for cluster quorum
Copiar o link

You can use an additional cluster node to serve as an extra quorum vote. In the following steps we configure such a majority-maker node dc3mm in the existing cluster.

6.2.1. Preparing the majority-maker node
Copiar o link

  • Install the node with the same operating system version as your HANA nodes.
  • Configure the RHEL High Availability repository on the majority-maker node.
  • If the firewalld service is installed and you are not using it on your hosts, disable the service. See Disabling the firewalld service.

6.2.2. Updating the host names in /etc/hosts
Copiar o link

As for all cluster nodes, we recommend that you also add the majority-maker cluster node to the /etc/hosts file on each node. On the new dc3mm node you add all cluster nodes.

Procedure

  • Add the new host to the /etc/hosts on all existing cluster nodes and add all hosts to the new dc3mm host, so that the resulting /etc/hosts entries are the same on all nodes: 

    [root]# cat /etc/hosts
...
192.168.100.101 dc1hana1.example.com dc1hana1
192.168.100.102 dc1hana2.example.com dc1hana2
192.168.100.103 dc2hana1.example.com dc2hana1
192.168.100.104 dc2hana2.example.com dc2hana2
192.168.100.110 dc3mm.example.com    dc3mm

Verification

  • Check that you can ping the hosts between each other. This step is optional and an example only for a basic verification. The system resolves entries in /etc/hosts when you use the ping command: 

    [root]# ping dc3mm.example.com
PING dc3mm.example.com (192.168.100.110) 56(84) bytes of data.
64 bytes from dc3mm.example.com (192.168.100.110): icmp_seq=1 ttl=64 time=0.017 ms
…

6.2.3. Updating the cluster clone resources
Copiar o link

The cluster automatically creates one copy of a clone resource for each cluster node and uses this to calculate resource allocations. For example, if the cluster consists of 4 HANA nodes and you add an additional majority-maker node for a quorum vote only, the cluster automatically calculates with all 5 cluster members for clone resource assignments. However, including this non-HANA node in the calculations can lead to unexpected impact when the cluster moves resources.

To prevent this influence you must configure the clone-max limit for all cloned resources explicitly to only the number of HANA nodes. Adjust the clone configuration before you add the new node to the cluster.

Procedure

  1. Update the SAPHanaTopology resource clone with a limit to the number of HANA nodes, for example, 4: 

    [root]# pcs resource update cln_SAPHanaTop_<SID>_HDB<instance> meta clone-max=4

  2. Update the SAPHanaController resource clone with a limit to the number of HANA nodes, for example, 4: 

    [root]# pcs resource update cln_SAPHanaCon_<SID>_HDB<instance> meta clone-max=4

  3. Optional: If you have configured the SAPHanaFilesystem resource clone, then also update it with a limit to the number of HANA nodes, for example, 4: 

    [root]# pcs resource update cln_SAPHanaFil_<SID>_HDB<instance> meta clone-max=4

Verification

  • Check that the clone-max option is correct for all clone resources: 

    [root]# pcs resource config | grep -i clone
Clone: cln_SAPHanaTop_RH1_HDB02
    clone-max=4
    clone-node-max=1
Clone: cln_SAPHanaCon_RH1_HDB02
    clone-max=4
    clone-node-max=1
Clone: cln_SAPHanaFil_RH1_HDB02
    clone-max=4
    clone-node-max=1
    • clone-max must be the number of HANA nodes, for example, 4. When this option is not displayed, it defaults to the total number of cluster nodes instead.

6.2.4. Installing the cluster components on the majority-maker node
Copiar o link

Install the same cluster packages as on the existing cluster nodes to prepare the host in the same way.

Prerequisites

  • You have configured the RHEL High Availability repository on the majority-maker host.

Procedure

  1. Install the Red Hat High Availability Add-On software packages from the High Availability repository. Choose the same fence agents as you are using on the existing cluster nodes: 

    [root]# dnf install pcs pacemaker fence-agents-<model>

  2. Start and enable the pcsd service on the new nodes. The --now parameter automatically starts the enabled service: 

    [root]# systemctl enable --now pcsd.service

  3. Optional: If you use the local firewalld service you must enable the ports that are required by the Red Hat High Availability Add-On. Run this on the new node: 

    [root]# firewall-cmd --add-service=high-availability
[root]# firewall-cmd --runtime-to-permanent

  4. Set a password for the user hacluster on the new node using the same password: 

    [root]# passwd hacluster

Verification

  1. Check that the pcsd service is running and shows as loaded and active on the new node: 

    [root]# systemctl status pcsd.service
● pcsd.service - PCS GUI and remote configuration interface
     Loaded: loaded (/usr/lib/systemd/system/pcsd.service; enabled; preset: disabled)
     Active: active (running) since …
…

6.2.5. Adding the new node to the cluster
Copiar o link

Add the dedicated majority-maker node as a regular cluster member.

Prerequisites

  • You have configured a cluster to which you want to add this node as a member.

Procedure

  1. Authenticate the user hacluster for the new node in the cluster. Run this on one cluster node: 

    [root]# pcs host auth dc3mm
Username: hacluster
Password:
dc3mm: Authorized
    • Enter the node name with or without FQDN, as defined in the /etc/hosts file.
    • Enter the hacluster user password in the prompt.

  2. Add the new node to the cluster. Run this on one cluster node: 

    [root]# pcs cluster node add dc3mm
No addresses specified for host 'dc3mm', using 'dc3mm'
Disabling sbd...
dc3mm: sbd disabled
Sending 'corosync authkey', 'pacemaker authkey' to 'dc3mm'
dc3mm: successful distribution of the file 'corosync authkey'
dc3mm: successful distribution of the file 'pacemaker authkey'
Sending updated corosync.conf to nodes...
dc2hana1: Succeeded
dc3mm: Succeeded
dc1hana2: Succeeded
dc2hana2: Succeeded
dc1hana1: Succeeded
dc2hana1: Corosync configuration reloaded

  3. Add a location constraint to prevent any HANA resource from running on this node, and also prevent the cluster from trying to check the initial resource status. The following constraint definition uses a regexp expression to match all HANA resources. If required, adjust the pattern to match your resource names: 

    [root]# pcs constraint location add avoid-dc3mm \
regexp%.*SAPHana.* dc3mm -- -INFINITY resource-discovery=never

  4. Enable the cluster on the new node to be started automatically on system start. Run on any node: 

    [root]# pcs cluster enable dc3mm
dc3mm: Cluster Enabled

  5. Start the cluster on the new node: 

    [root]# pcs cluster start dc3mm
dc3mm: Starting Cluster...

Verification

  1. Check the location constraint that keeps HANA resources off the new node: 

    [root]# pcs constraint location --full
Location Constraints:
…
  resource pattern '.*SAPHana.*' avoids node 'dc3mm' with score INFINITY (id: avoid-dc3mm)
    resource-discovery=never

  2. Check the cluster status. Verify that the cluster daemon services are in the desired state. Run this on the new node to also verify the local daemon status at the end: 

    [root]# pcs status --full
…

Node List:
  * Node dc1hana1 (1): online, feature set 3.19.6
  * Node dc1hana2 (2): online, feature set 3.19.6
  * Node dc2hana1 (3): online, feature set 3.19.6
  * Node dc2hana2 (4): online, feature set 3.19.6
  * Node dc3mm (5): online, feature set 3.19.6

Full List of Resources:
  * Clone Set: cln_SAPHanaTop_RH1_HDB02 [rsc_SAPHanaTop_RH1_HDB02]:
    * rsc_SAPHanaTop_RH1_HDB02  (ocf:heartbeat:SAPHanaTopology):         Started dc1hana2
    * rsc_SAPHanaTop_RH1_HDB02  (ocf:heartbeat:SAPHanaTopology):         Started dc1hana1
    * rsc_SAPHanaTop_RH1_HDB02  (ocf:heartbeat:SAPHanaTopology):         Started dc2hana1
    * rsc_SAPHanaTop_RH1_HDB02  (ocf:heartbeat:SAPHanaTopology):         Started dc2hana2
  * Clone Set: cln_SAPHanaCon_RH1_HDB02 [rsc_SAPHanaCon_RH1_HDB02] (promotable):
    * rsc_SAPHanaCon_RH1_HDB02  (ocf:heartbeat:SAPHanaController):       Unpromoted dc1hana2
    * rsc_SAPHanaCon_RH1_HDB02  (ocf:heartbeat:SAPHanaController):       Promoted dc1hana1
    * rsc_SAPHanaCon_RH1_HDB02  (ocf:heartbeat:SAPHanaController):       Unpromoted dc2hana1
    * rsc_SAPHanaCon_RH1_HDB02  (ocf:heartbeat:SAPHanaController):       Unpromoted dc2hana2
  * Clone Set: cln_SAPHanaFil_RH1_HDB02 [rsc_SAPHanaFil_RH1_HDB02]:
    * rsc_SAPHanaFil_RH1_HDB02  (ocf:heartbeat:SAPHanaFilesystem):       Started dc1hana2
    * rsc_SAPHanaFil_RH1_HDB02  (ocf:heartbeat:SAPHanaFilesystem):       Started dc1hana1
    * rsc_SAPHanaFil_RH1_HDB02  (ocf:heartbeat:SAPHanaFilesystem):       Started dc2hana1
    * rsc_SAPHanaFil_RH1_HDB02  (ocf:heartbeat:SAPHanaFilesystem):       Started dc2hana2
  * rsc_vip_RH1_HDB02_primary   (ocf:heartbeat:IPAddr2):         Started dc1hana1
  * rsc_vip_RH1_HDB02_readonly  (ocf:heartbeat:IPAddr2):         Started dc2hana1

…

PCSD Status:
  dc1hana1: Online
  dc1hana2: Online
  dc2hana1: Online
  dc2hana2: Online
  dc3mm: Online

Daemon Status:
  corosync: active/enabled
  pacemaker: active/enabled
  pcsd: active/enabled
    • The new node must be displayed in the Node List and in the PCSD Status list.
    • The new node must not show a status in the Full List of Resources.

  3. Verify the quorum status. The new node adds a vote, so that there is now an odd number of votes and there can be no equal 50/50 split anymore: 

    [root]# pcs quorum status
Quorum information
------------------
Date:             Tue Sep  2 14:56:29 2025
Quorum provider:  corosync_votequorum
Nodes:            5
Node ID:          5
Ring ID:          1.27
Quorate:          Yes

Votequorum information
----------------------
Expected votes:   5
Highest expected: 5
Total votes:      5
Quorum:           3
Flags:            Quorate

Membership information
----------------------
    Nodeid      Votes    Qdevice Name
         1          1         NR dc1hana1
         2          1         NR dc1hana2
         3          1         NR dc2hana1
         4          1         NR dc2hana2
         5          1         NR dc3mm (local)

Next steps

Red Hat logoGithubredditYoutubeTwitter

Aprender

Experimente, compre e venda

Comunidades

Sobre a documentação da Red Hat

Ajudamos os usuários da Red Hat a inovar e atingir seus objetivos com nossos produtos e serviços com conteúdo em que podem confiar. Explore nossas atualizações recentes.

Tornando o open source mais inclusivo

A Red Hat está comprometida em substituir a linguagem problemática em nosso código, documentação e propriedades da web. Para mais detalhes veja o Blog da Red Hat.

Sobre a Red Hat

Fornecemos soluções robustas que facilitam o trabalho das empresas em plataformas e ambientes, desde o data center principal até a borda da rede.

Theme

© 2026 Red HatVoltar ao topo