High availability of SAP HANA scale-up with Azure NetApp Files on RHEL
This article describes how to configure SAP HANA System Replication in scale-up deployment, when the HANA file systems are mounted via NFS, by using Azure NetApp Files. In the example configurations and installation commands, instance number 03 and HANA System ID HN1 are used. SAP HANA System Replication consists of one primary node and at least one secondary node.
When steps in this document are marked with the following prefixes, the meaning is as follows:
- [A]: The step applies to all nodes
- [1]: The step applies to node1 only
- [2]: The step applies to node2 only
Prerequisites
Read the following SAP Notes and papers first:
- SAP Note 1928533, which has:
- The list of Azure virtual machine (VM) sizes that are supported for the deployment of SAP software.
- Important capacity information for Azure VM sizes.
- The supported SAP software and operating system (OS) and database combinations.
- The required SAP kernel version for Windows and Linux on Microsoft Azure.
- SAP Note 2015553 lists prerequisites for SAP-supported SAP software deployments in Azure.
- SAP Note 405827 lists recommended file systems for HANA environments.
- SAP Note 2002167 has recommended OS settings for Red Hat Enterprise Linux.
- SAP Note 2009879 has SAP HANA Guidelines for Red Hat Enterprise Linux.
- SAP Note 3108302 has SAP HANA Guidelines for Red Hat Enterprise Linux 9.x.
- SAP Note 2178632 has detailed information about all monitoring metrics reported for SAP in Azure.
- SAP Note 2191498 has the required SAP Host Agent version for Linux in Azure.
- SAP Note 2243692 has information about SAP licensing on Linux in Azure.
- SAP Note 1999351 has more troubleshooting information for the Azure Enhanced Monitoring Extension for SAP.
- SAP Community Wiki has all required SAP Notes for Linux.
- Azure Virtual Machines planning and implementation for SAP on Linux
- Azure Virtual Machines deployment for SAP on Linux
- Azure Virtual Machines DBMS deployment for SAP on Linux
- SAP HANA system replication in Pacemaker cluster
- General Red Hat Enterprise Linux (RHEL) documentation:
- Azure-specific RHEL documentation:
- Support Policies for RHEL High Availability Clusters - Microsoft Azure Virtual Machines as Cluster Members
- Installing and Configuring a Red Hat Enterprise Linux 7.4 (and later) High-Availability Cluster on Microsoft Azure
- Configure SAP HANA scale-up system replication in a Pacemaker cluster when the HANA file systems are on NFS shares
- NFS v4.1 volumes on Azure NetApp Files for SAP HANA
Overview
Traditionally in a scale-up environment, all file systems for SAP HANA are mounted from local storage. Setting up high availability (HA) of SAP HANA System Replication on Red Hat Enterprise Linux is published in Set up SAP HANA System Replication on RHEL.
To achieve SAP HANA HA of a scale-up system on Azure NetApp Files NFS shares, we need some more resource configuration in the cluster, in order for HANA resources to recover, when one node loses access to the NFS shares on Azure NetApp Files. The cluster manages the NFS mounts, allowing it to monitor the health of the resources. The dependencies between the file system mounts and the SAP HANA resources are enforced.
.
SAP HANA file systems are mounted on NFS shares by using Azure NetApp Files on each node. File systems /hana/data
, /hana/log
, and /hana/shared
are unique to each node.
Mounted on node1 (hanadb1):
- 10.32.2.4:/hanadb1-data-mnt00001 on /hana/data
- 10.32.2.4:/hanadb1-log-mnt00001 on /hana/log
- 10.32.2.4:/hanadb1-shared-mnt00001 on /hana/shared
Mounted on node2 (hanadb2):
- 10.32.2.4:/hanadb2-data-mnt00001 on /hana/data
- 10.32.2.4:/hanadb2-log-mnt00001 on /hana/log
- 10.32.2.4:/hanadb2-shared-mnt00001 on /hana/shared
Note
File systems /hana/shared
, /hana/data
, and /hana/log
aren't shared between the two nodes. Each cluster node has its own separate file systems.
The SAP HANA System Replication configuration uses a dedicated virtual hostname and virtual IP addresses. On Azure, a load balancer is required to use a virtual IP address. The configuration shown here has a load balancer with:
- Front-end IP address: 10.32.0.10 for hn1-db
- Probe port: 62503
Set up the Azure NetApp Files infrastructure
Before you proceed with the setup for Azure NetApp Files infrastructure, familiarize yourself with the Azure NetApp Files documentation.
Azure NetApp Files is available in several Azure regions. Check to see whether your selected Azure region offers Azure NetApp Files.
For information about the availability of Azure NetApp Files by Azure region, see Azure NetApp Files availability by Azure region.
Important considerations
As you're creating your Azure NetApp Files volumes for SAP HANA scale-up systems, be aware of the important considerations documented in NFS v4.1 volumes on Azure NetApp Files for SAP HANA.
Sizing of HANA database on Azure NetApp Files
The throughput of an Azure NetApp Files volume is a function of the volume size and service level, as documented in Service level for Azure NetApp Files.
While you're designing the infrastructure for SAP HANA on Azure with Azure NetApp Files, be aware of the recommendations in NFS v4.1 volumes on Azure NetApp Files for SAP HANA.
The configuration in this article is presented with simple Azure NetApp Files volumes.
Important
For production systems, where performance is a key, we recommend that you evaluate and consider using Azure NetApp Files application volume group for SAP HANA.
Deploy Azure NetApp Files resources
The following instructions assume that you already deployed your Azure virtual network. The Azure NetApp Files resources and VMs, where the Azure NetApp Files resources will be mounted, must be deployed in the same Azure virtual network or in peered Azure virtual networks.
Create a NetApp account in your selected Azure region by following the instructions in Create a NetApp account.
Set up an Azure NetApp Files capacity pool by following the instructions in Set up an Azure NetApp Files capacity pool.
The HANA architecture shown in this article uses a single Azure NetApp Files capacity pool at the Ultra service level. For HANA workloads on Azure, we recommend using an Azure NetApp Files Ultra or Premium service Level.
Delegate a subnet to Azure NetApp Files, as described in the instructions in Delegate a subnet to Azure NetApp Files.
Deploy Azure NetApp Files volumes by following the instructions in Create an NFS volume for Azure NetApp Files.
As you're deploying the volumes, be sure to select the NFSv4.1 version. Deploy the volumes in the designated Azure NetApp Files subnet. The IP addresses of the Azure NetApp volumes are assigned automatically.
Keep in mind that the Azure NetApp Files resources and the Azure VMs must be in the same Azure virtual network or in peered Azure virtual networks. For example,
hanadb1-data-mnt00001
andhanadb1-log-mnt00001
are the volume names andnfs://10.32.2.4/hanadb1-data-mnt00001
andnfs://10.32.2.4/hanadb1-log-mnt00001
are the file paths for the Azure NetApp Files volumes.On hanadb1:
- Volume hanadb1-data-mnt00001 (nfs://10.32.2.4:/hanadb1-data-mnt00001)
- Volume hanadb1-log-mnt00001 (nfs://10.32.2.4:/hanadb1-log-mnt00001)
- Volume hanadb1-shared-mnt00001 (nfs://10.32.2.4:/hanadb1-shared-mnt00001)
On hanadb2:
- Volume hanadb2-data-mnt00001 (nfs://10.32.2.4:/hanadb2-data-mnt00001)
- Volume hanadb2-log-mnt00001 (nfs://10.32.2.4:/hanadb2-log-mnt00001)
- Volume hanadb2-shared-mnt00001 (nfs://10.32.2.4:/hanadb2-shared-mnt00001)
Note
All commands to mount /hana/shared
in this article are presented for NFSv4.1 /hana/shared
volumes.
If you deployed the /hana/shared
volumes as NFSv3 volumes, don't forget to adjust the mount commands for /hana/shared
for NFSv3.
Prepare the infrastructure
Azure Marketplace contains images qualified for SAP HANA with the High Availability add-on, which you can use to deploy new VMs by using various versions of Red Hat.
Deploy Linux VMs manually via the Azure portal
This document assumes that you've already deployed a resource group, an Azure virtual network, and a subnet.
Deploy VMs for SAP HANA. Choose a suitable RHEL image that's supported for the HANA system. You can deploy a VM in any one of the availability options: virtual machine scale set, availability zone, or availability set.
Important
Make sure that the OS you select is SAP certified for SAP HANA on the specific VM types that you plan to use in your deployment. You can look up SAP HANA-certified VM types and their OS releases in SAP HANA Certified IaaS Platforms. Make sure that you look at the details of the VM type to get the complete list of SAP HANA-supported OS releases for the specific VM type.
Configure Azure load balancer
During VM configuration, you have an option to create or select exiting load balancer in networking section. Follow below steps, to setup standard load balancer for high availability setup of HANA database.
Follow the steps in Create load balancer to set up a standard load balancer for a high-availability SAP system by using the Azure portal. During the setup of the load balancer, consider the following points:
- Frontend IP Configuration: Create a front-end IP. Select the same virtual network and subnet name as your database virtual machines.
- Backend Pool: Create a back-end pool and add database VMs.
- Inbound rules: Create a load-balancing rule. Follow the same steps for both load-balancing rules.
- Frontend IP address: Select a front-end IP.
- Backend pool: Select a back-end pool.
- High-availability ports: Select this option.
- Protocol: Select TCP.
- Health Probe: Create a health probe with the following details:
- Protocol: Select TCP.
- Port: For example, 625<instance-no.>.
- Interval: Enter 5.
- Probe Threshold: Enter 2.
- Idle timeout (minutes): Enter 30.
- Enable Floating IP: Select this option.
Note
The health probe configuration property numberOfProbes
, otherwise known as Unhealthy threshold in the portal, isn't respected. To control the number of successful or failed consecutive probes, set the property probeThreshold
to 2
. It's currently not possible to set this property by using the Azure portal, so use either the Azure CLI or the PowerShell command.
For more information about the required ports for SAP HANA, read the chapter Connections to Tenant Databases in the SAP HANA Tenant Databases guide or SAP Note 2388694.
Note
When VMs without public IP addresses are placed in the back-end pool of an internal (no public IP address) instance of Standard Azure Load Balancer, there's no outbound internet connectivity, unless more configuration is performed to allow routing to public endpoints. For more information on how to achieve outbound connectivity, see Public endpoint connectivity for virtual machines using Standard Azure Load Balancer in SAP high-availability scenarios.
Important
Don't enable TCP timestamps on Azure VMs placed behind Azure Load Balancer. Enabling TCP timestamps could cause the health probes to fail. Set the parameter net.ipv4.tcp_timestamps to 0. For more information, see Load Balancer health probes and SAP Note 2382421.
Mount the Azure NetApp Files volume
[A] Create mount points for the HANA database volumes.
sudo mkdir -p /hana/data sudo mkdir -p /hana/log sudo mkdir -p /hana/shared
[A] Verify the NFS domain setting. Make sure that the domain is configured as the default Azure NetApp Files domain, that is, defaultv4iddomain.com, and the mapping is set to nobody.
sudo cat /etc/idmapd.conf
Example output:
[General] Domain = defaultv4iddomain.com [Mapping] Nobody-User = nobody Nobody-Group = nobody
Important
Make sure to set the NFS domain in
/etc/idmapd.conf
on the VM to match the default domain configuration on Azure NetApp Files: defaultv4iddomain.com. If there's a mismatch between the domain configuration on the NFS client (that is, the VM) and the NFS server (that is, the Azure NetApp Files configuration), then the permissions for files on Azure NetApp Files volumes that are mounted on the VMs display asnobody
.[1] Mount the node-specific volumes on node1 (hanadb1).
sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb1-shared-mnt00001 /hana/shared sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb1-log-mnt00001 /hana/log sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb1-data-mnt00001 /hana/data
[2] Mount the node-specific volumes on node2 (hanadb2).
sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb2-shared-mnt00001 /hana/shared sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb2-log-mnt00001 /hana/log sudo mount -o rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys 10.32.2.4:/hanadb2-data-mnt00001 /hana/data
[A] Verify that all HANA volumes are mounted with NFS protocol version NFSv4.
sudo nfsstat -m
Verify that the flag
vers
is set to 4.1. Example from hanadb1:/hana/log from 10.32.2.4:/hanadb1-log-mnt00001 Flags: rw,noatime,vers=4.1,rsize=262144,wsize=262144,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=10.32.0.4,local_lock=none,addr=10.32.2.4 /hana/data from 10.32.2.4:/hanadb1-data-mnt00001 Flags: rw,noatime,vers=4.1,rsize=262144,wsize=262144,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=10.32.0.4,local_lock=none,addr=10.32.2.4 /hana/shared from 10.32.2.4:/hanadb1-shared-mnt00001 Flags: rw,noatime,vers=4.1,rsize=262144,wsize=262144,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=10.32.0.4,local_lock=none,addr=10.32.2.4
[A] Verify nfs4_disable_idmapping. It should be set to Y. To create the directory structure where nfs4_disable_idmapping is located, run the mount command. You can't manually create the directory under
/sys/modules
because access is reserved for the kernel and drivers.Check
nfs4_disable_idmapping
.sudo cat /sys/module/nfs/parameters/nfs4_disable_idmapping
If you need to set
nfs4_disable_idmapping
to:sudo echo "Y" > /sys/module/nfs/parameters/nfs4_disable_idmapping
Make the configuration permanent.
sudo echo "options nfs nfs4_disable_idmapping=Y" >> /etc/modprobe.d/nfs.conf
For more information on how to change the
nfs_disable_idmapping
parameter, see the Red Hat Knowledge Base.
SAP HANA installation
[A] Set up hostname resolution for all hosts.
You can either use a DNS server or modify the
/etc/hosts
file on all nodes. This example shows you how to use the/etc/hosts
file. Replace the IP address and the hostname in the following commands:sudo vi /etc/hosts
Insert the following lines in the
/etc/hosts
file. Change the IP address and hostname to match your environment.10.32.0.4 hanadb1 10.32.0.5 hanadb2
[A] Prepare the OS for running SAP HANA on Azure NetApp with NFS, as described in SAP Note 3024346 - Linux Kernel Settings for NetApp NFS. Create configuration file
/etc/sysctl.d/91-NetApp-HANA.conf
for the NetApp configuration settings.sudo vi /etc/sysctl.d/91-NetApp-HANA.conf
Add the following entries in the configuration file.
net.core.rmem_max = 16777216 net.core.wmem_max = 16777216 net.ipv4.tcp_rmem = 4096 131072 16777216 net.ipv4.tcp_wmem = 4096 16384 16777216 net.core.netdev_max_backlog = 300000 net.ipv4.tcp_slow_start_after_idle=0 net.ipv4.tcp_no_metrics_save = 1 net.ipv4.tcp_moderate_rcvbuf = 1 net.ipv4.tcp_window_scaling = 1 net.ipv4.tcp_sack = 1
[A] Create the configuration file
/etc/sysctl.d/ms-az.conf
with more optimization settings.sudo vi /etc/sysctl.d/ms-az.conf
Add the following entries in the configuration file.
net.ipv6.conf.all.disable_ipv6 = 1 net.ipv4.tcp_max_syn_backlog = 16348 net.ipv4.conf.all.rp_filter = 0 sunrpc.tcp_slot_table_entries = 128 vm.swappiness=10
Tip
Avoid setting
net.ipv4.ip_local_port_range
andnet.ipv4.ip_local_reserved_ports
explicitly in thesysctl
configuration files to allow the SAP Host Agent to manage the port ranges. For more information, see SAP Note 2382421.[A] Adjust the
sunrpc
settings, as recommended in SAP Note 3024346 - Linux Kernel Settings for NetApp NFS.sudo vi /etc/modprobe.d/sunrpc.conf
Insert the following line:
options sunrpc tcp_max_slot_table_entries=128
[A] Perform RHEL OS configuration for HANA.
Configure the OS as described in the following SAP Notes based on your RHEL version:
- 2292690 - SAP HANA DB: Recommended OS settings for RHEL 7
- 2777782 - SAP HANA DB: Recommended OS Settings for RHEL 8
- 2455582 - Linux: Running SAP applications compiled with GCC 6.x
- 2593824 - Linux: Running SAP applications compiled with GCC 7.x
- 2886607 - Linux: Running SAP applications compiled with GCC 9.x
[A] Install the SAP HANA.
Starting with HANA 2.0 SPS 01, MDC is the default option. When you install the HANA system, SYSTEMDB and a tenant with the same SID are created together. In some cases, you don't want the default tenant. If you don't want to create an initial tenant along with the installation, you can follow SAP Note 2629711.
Run the hdblcm program from the HANA DVD. Enter the following values at the prompt:
- Choose installation: Enter 1 (for install).
- Select more components for installation: Enter 1.
- Enter Installation Path [/hana/shared]: Select Enter to accept the default.
- Enter Local Host Name [..]: Select Enter to accept the default. Do you want to add additional hosts to the system? (y/n) [n]: n.
- Enter SAP HANA System ID: Enter HN1.
- Enter Instance Number [00]: Enter 03.
- Select Database Mode / Enter Index [1]: Select Enter to accept the default.
- Select System Usage / Enter Index [4]: Enter 4 (for custom).
- Enter Location of Data Volumes [/hana/data]: Select Enter to accept the default.
- Enter Location of Log Volumes [/hana/log]: Select Enter to accept the default.
- Restrict maximum memory allocation? [n]: Select Enter to accept the default.
- Enter Certificate Host Name For Host '...' [...]: Select Enter to accept the default.
- Enter SAP Host Agent User (sapadm) Password: Enter the host agent user password.
- Confirm SAP Host Agent User (sapadm) Password: Enter the host agent user password again to confirm.
- Enter System Administrator (hn1adm) Password: Enter the system administrator password.
- Confirm System Administrator (hn1adm) Password: Enter the system administrator password again to confirm.
- Enter System Administrator Home Directory [/usr/sap/HN1/home]: Select Enter to accept the default.
- Enter System Administrator Login Shell [/bin/sh]: Select Enter to accept the default.
- Enter System Administrator User ID [1001]: Select Enter to accept the default.
- Enter ID of User Group (sapsys) [79]: Select Enter to accept the default.
- Enter Database User (SYSTEM) Password: Enter the database user password.
- Confirm Database User (SYSTEM) Password: Enter the database user password again to confirm.
- Restart system after machine reboot? [n]: Select Enter to accept the default.
- Do you want to continue? (y/n): Validate the summary. Enter y to continue.
[A] Upgrade the SAP Host Agent.
Download the latest SAP Host Agent archive from the SAP Software Center and run the following command to upgrade the agent. Replace the path to the archive to point to the file that you downloaded:
sudo /usr/sap/hostctrl/exe/saphostexec -upgrade -archive <path to SAP Host Agent SAR>
[A] Configure a firewall.
Create the firewall rule for the Azure Load Balancer probe port.
sudo firewall-cmd --zone=public --add-port=62503/tcp sudo firewall-cmd --zone=public --add-port=62503/tcp –permanent
Configure SAP HANA System Replication
Follow the steps in Set up SAP HANA System Replication to configure SAP HANA System Replication.
Cluster configuration
This section describes the steps required for a cluster to operate seamlessly when SAP HANA is installed on NFS shares by using Azure NetApp Files.
Create a Pacemaker cluster
Follow the steps in Set up Pacemaker on Red Hat Enterprise Linux in Azure to create a basic Pacemaker cluster for this HANA server.
Important
With the systemd based SAP Startup Framework, SAP HANA instances can now be managed by systemd. The minimum required Red Hat Enterprise Linux (RHEL) version is RHEL 8 for SAP. As outlined in SAP Note 3189534, any new installations of SAP HANA SPS07 revision 70 or above, or updates to HANA systems to HANA 2.0 SPS07 revision 70 or above, SAP Startup framework will be automatically registered with systemd.
When using HA solutions to manage SAP HANA system replication in combination with systemd-enabled SAP HANA instances (refer to SAP Note 3189534), additional steps are necessary to ensure that the HA cluster can manage the SAP instance without systemd interference. So, for SAP HANA system integrated with systemd, additional steps outlined in Red Hat KBA 7029705 must be followed on all cluster nodes.
Implement the Python system replication hook SAPHanaSR
This step is an important one to optimize the integration with the cluster and improve the detection when a cluster failover is needed. We highly recommend that you configure the SAPHanaSR Python hook. Follow the steps in Implement the Python system replication hook SAPHanaSR.
Configure file system resources
In this example, each cluster node has its own HANA NFS file systems /hana/shared
, /hana/data
, and /hana/log
.
[1] Put the cluster in maintenance mode.
sudo pcs property set maintenance-mode=true
[1] Create the file system resources for the hanadb1 mounts.
sudo pcs resource create hana_data1 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb1-data-mnt00001 directory=/hana/data fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb1_nfs sudo pcs resource create hana_log1 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb1-log-mnt00001 directory=/hana/log fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb1_nfs sudo pcs resource create hana_shared1 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb1-shared-mnt00001 directory=/hana/shared fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb1_nfs
[2] Create the file system resources for the hanadb2 mounts.
sudo pcs resource create hana_data2 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb2-data-mnt00001 directory=/hana/data fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb2_nfs sudo pcs resource create hana_log2 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb2-log-mnt00001 directory=/hana/log fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb2_nfs sudo pcs resource create hana_shared2 ocf:heartbeat:Filesystem device=10.32.2.4:/hanadb2-shared-mnt00001 directory=/hana/shared fstype=nfs options=rw,nfsvers=4.1,hard,timeo=600,rsize=262144,wsize=262144,noatime,lock,_netdev,sec=sys op monitor interval=20s on-fail=fence timeout=120s OCF_CHECK_LEVEL=20 --group hanadb2_nfs
The
OCF_CHECK_LEVEL=20
attribute is added to the monitor operation so that each monitor performs a read/write test on the file system. Without this attribute, the monitor operation only verifies that the file system is mounted. This can be a problem because when connectivity is lost, the file system might remain mounted despite being inaccessible.The
on-fail=fence
attribute is also added to the monitor operation. With this option, if the monitor operation fails on a node, that node is immediately fenced. Without this option, the default behavior is to stop all resources that depend on the failed resource, restart the failed resource, and then start all the resources that depend on the failed resource.Not only can this behavior take a long time when an SAPHana resource depends on the failed resource, but it also can fail altogether. The SAPHana resource can't stop successfully if the NFS server holding the HANA executables is inaccessible.
The suggested timeout values allow the cluster resources to withstand protocol-specific pause, related to NFSv4.1 lease renewals. For more information, see NFS in NetApp Best practice. The timeouts in the preceding configuration might need to be adapted to the specific SAP setup.
For workloads that require higher throughput, consider using the
nconnect
mount option, as described in NFS v4.1 volumes on Azure NetApp Files for SAP HANA. Check ifnconnect
is supported by Azure NetApp Files on your Linux release.[1] Configure location constraints.
Configure location constraints to ensure that the resources that manage hanadb1 unique mounts can never run on hanadb2, and vice versa.
sudo pcs constraint location hanadb1_nfs rule score=-INFINITY resource-discovery=never \#uname eq hanadb2 sudo pcs constraint location hanadb2_nfs rule score=-INFINITY resource-discovery=never \#uname eq hanadb1
The
resource-discovery=never
option is set because the unique mounts for each node share the same mount point. For example,hana_data1
uses mount point/hana/data
, andhana_data2
also uses mount point/hana/data
. Sharing the same mount point can cause a false positive for a probe operation, when resource state is checked at cluster startup, and it can in turn cause unnecessary recovery behavior. To avoid this scenario, setresource-discovery=never
.[1] Configure attribute resources.
Configure attribute resources. These attributes are set to true if all of a node's NFS mounts (
/hana/data
,/hana/log
, and/hana/data
) are mounted. Otherwise, they're set to false.sudo pcs resource create hana_nfs1_active ocf:pacemaker:attribute active_value=true inactive_value=false name=hana_nfs1_active sudo pcs resource create hana_nfs2_active ocf:pacemaker:attribute active_value=true inactive_value=false name=hana_nfs2_active
[1] Configure location constraints.
Configure location constraints to ensure that hanadb1's attribute resource never runs on hanadb2, and vice versa.
sudo pcs constraint location hana_nfs1_active avoids hanadb2 sudo pcs constraint location hana_nfs2_active avoids hanadb1
[1] Create ordering constraints.
Configure ordering constraints so that a node's attribute resources start only after all of the node's NFS mounts are mounted.
sudo pcs constraint order hanadb1_nfs then hana_nfs1_active sudo pcs constraint order hanadb2_nfs then hana_nfs2_active
Tip
If your configuration includes file systems, outside of group
hanadb1_nfs
orhanadb2_nfs
, include thesequential=false
option so that there are no ordering dependencies among the file systems. All file systems must start beforehana_nfs1_active
, but they don't need to start in any order relative to each other. For more information, see How do I configure SAP HANA System Replication in Scale-Up in a Pacemaker cluster when the HANA file systems are on NFS shares
Configure SAP HANA cluster resources
Follow the steps in Create SAP HANA cluster resources to create the SAP HANA resources in the cluster. After SAP HANA resources are created, you need to create a location rule constraint between SAP HANA resources and file systems (NFS mounts).
[1] Configure constraints between the SAP HANA resources and the NFS mounts.
Location rule constraints are set so that the SAP HANA resources can run on a node only if all of the node's NFS mounts are mounted.
sudo pcs constraint location SAPHanaTopology_HN1_03-clone rule score=-INFINITY hana_nfs1_active ne true and hana_nfs2_active ne true
On RHEL 7.x:
sudo pcs constraint location SAPHana_HN1_03-master rule score=-INFINITY hana_nfs1_active ne true and hana_nfs2_active ne true
On RHEL 8.x/9.x:
sudo pcs constraint location SAPHana_HN1_03-clone rule score=-INFINITY hana_nfs1_active ne true and hana_nfs2_active ne true
[1] Configure ordering constraints so that the SAP resources on a node will stop ahead of a stop for any of the NFS mounts.
pcs constraint order stop SAPHanaTopology_HN1_03-clone then stop hanadb1_nfs pcs constraint order stop SAPHanaTopology_HN1_03-clone then stop hanadb2_nfs
On RHEL 7.x:
pcs constraint order stop SAPHana_HN1_03-master then stop hanadb1_nfs pcs constraint order stop SAPHana_HN1_03-master then stop hanadb2_nfs
On RHEL 8.x/9.x:
pcs constraint order stop SAPHana_HN1_03-clone then stop hanadb1_nfs pcs constraint order stop SAPHana_HN1_03-clone then stop hanadb2_nfs
Take the cluster out of maintenance mode.
sudo pcs property set maintenance-mode=false
Check the status of the cluster and all the resources.
Note
This article contains references to a term that Microsoft no longer uses. When the term is removed from the software, we'll remove it from this article.
sudo pcs status
Example output:
Online: [ hanadb1 hanadb2 ] Full list of resources: rsc_hdb_azr_agt(stonith:fence_azure_arm): Started hanadb1 Resource Group: hanadb1_nfs hana_data1 (ocf::heartbeat:Filesystem):Started hanadb1 hana_log1 (ocf::heartbeat:Filesystem):Started hanadb1 hana_shared1 (ocf::heartbeat:Filesystem):Started hanadb1 Resource Group: hanadb2_nfs hana_data2 (ocf::heartbeat:Filesystem):Started hanadb2 hana_log2 (ocf::heartbeat:Filesystem):Started hanadb2 hana_shared2 (ocf::heartbeat:Filesystem):Started hanadb2 hana_nfs1_active (ocf::pacemaker:attribute): Started hanadb1 hana_nfs2_active (ocf::pacemaker:attribute): Started hanadb2 Clone Set: SAPHanaTopology_HN1_03-clone [SAPHanaTopology_HN1_03] Started: [ hanadb1 hanadb2 ] Master/Slave Set: SAPHana_HN1_03-master [SAPHana_HN1_03] Masters: [ hanadb1 ] Slaves: [ hanadb2 ] Resource Group: g_ip_HN1_03 nc_HN1_03 (ocf::heartbeat:azure-lb): Started hanadb1 vip_HN1_03 (ocf::heartbeat:IPaddr2): Started hanadb1
Configure HANA active/read-enabled system replication in Pacemaker cluster
Starting with SAP HANA 2.0 SPS 01, SAP allows active/read-enabled setups for SAP HANA System Replication, where the secondary systems of SAP HANA System Replication can be used actively for read-intense workloads. To support such a setup in a cluster, a second virtual IP address is required, which allows clients to access the secondary read-enabled SAP HANA database.
To ensure that the secondary replication site can still be accessed after a takeover has occurred, the cluster needs to move the virtual IP address around with the secondary of the SAPHana resource.
The extra configuration, which is required to manage HANA active/read-enabled System Replication in a Red Hat HA cluster with a second virtual IP, is described in Configure HANA Active/Read-Enabled System Replication in Pacemaker cluster.
Before you proceed further, make sure you've fully configured Red Hat High Availability Cluster managing SAP HANA database as described in the preceding sections of the documentation.
Test the cluster setup
This section describes how you can test your setup.
Before you start a test, make sure that Pacemaker doesn't have any failed action (via pcs status), there are no unexpected location constraints (for example, leftovers of a migration test), and that HANA system replication is in sync state, for example, with
systemReplicationStatus
:sudo su - hn1adm -c "python /usr/sap/HN1/HDB03/exe/python_support/systemReplicationStatus.py"
Verify the cluster configuration for a failure scenario when a node loses access to the NFS share (
/hana/shared
).The SAP HANA resource agents depend on binaries stored on
/hana/shared
to perform operations during failover. File system/hana/shared
is mounted over NFS in the presented scenario.It's difficult to simulate a failure where one of the servers loses access to the NFS share. As a test, you can remount the file system as read-only. This approach validates that the cluster can fail over, if access to
/hana/shared
is lost on the active node.Expected result: On making
/hana/shared
as a read-only file system, theOCF_CHECK_LEVEL
attribute of the resourcehana_shared1
, which performs read/write operations on file systems, fails. It isn't able to write anything on the file system and performs HANA resource failover. The same result is expected when your HANA node loses access to the NFS shares.Resource state before starting the test:
sudo pcs status
Example output:
Full list of resources: rsc_hdb_azr_agt (stonith:fence_azure_arm): Started hanadb1 Resource Group: hanadb1_nfs hana_data1 (ocf::heartbeat:Filesystem): Started hanadb1 hana_log1 (ocf::heartbeat:Filesystem): Started hanadb1 hana_shared1 (ocf::heartbeat:Filesystem): Started hanadb1 Resource Group: hanadb2_nfs hana_data2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_log2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_shared2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_nfs1_active (ocf::pacemaker:attribute): Started hanadb1 hana_nfs2_active (ocf::pacemaker:attribute): Started hanadb2 Clone Set: SAPHanaTopology_HN1_03-clone [SAPHanaTopology_HN1_03] Started: [ hanadb1 hanadb2 ] Master/Slave Set: SAPHana_HN1_03-master [SAPHana_HN1_03] Masters: [ hanadb1 ] Slaves: [ hanadb2 ] Resource Group: g_ip_HN1_03 nc_HN1_03 (ocf::heartbeat:azure-lb): Started hanadb1 vip_HN1_03 (ocf::heartbeat:IPaddr2): Started hanadb1
You can place
/hana/shared
in read-only mode on the active cluster node by using this command:sudo mount -o ro 10.32.2.4:/hanadb1-shared-mnt00001 /hana/shared
hanadb
will either reboot or power off based on the action set onstonith
(pcs property show stonith-action
). Once the server (hanadb1
) is down, the HANA resource moves tohanadb2
. You can check the status of the cluster fromhanadb2
.sudo pcs status
Example output:
Full list of resources: rsc_hdb_azr_agt (stonith:fence_azure_arm): Started hanadb2 Resource Group: hanadb1_nfs hana_data1 (ocf::heartbeat:Filesystem): Stopped hana_log1 (ocf::heartbeat:Filesystem): Stopped hana_shared1 (ocf::heartbeat:Filesystem): Stopped Resource Group: hanadb2_nfs hana_data2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_log2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_shared2 (ocf::heartbeat:Filesystem): Started hanadb2 hana_nfs1_active (ocf::pacemaker:attribute): Stopped hana_nfs2_active (ocf::pacemaker:attribute): Started hanadb2 Clone Set: SAPHanaTopology_HN1_03-clone [SAPHanaTopology_HN1_03] Started: [ hanadb2 ] Stopped: [ hanadb1 ] Master/Slave Set: SAPHana_HN1_03-master [SAPHana_HN1_03] Masters: [ hanadb2 ] Stopped: [ hanadb1 ] Resource Group: g_ip_HN1_03 nc_HN1_03 (ocf::heartbeat:azure-lb): Started hanadb2 vip_HN1_03 (ocf::heartbeat:IPaddr2): Started hanadb2
We recommend that you thoroughly test the SAP HANA cluster configuration by also performing the tests described in Set up SAP HANA System Replication on RHEL.