High availability for SAP NW 7.5 + SLES for SAP 15

Hi,

I am new to SUSE and HA. Although have working experience on Solaris.

Recently my SAP team wants to setup 2 node cluster for SAP NW 7.5 on SUSE for SAP in virtualized VMware environment for fail over of SAP ASCS and Database ASE.

I dont know how to do it.

Can someone pls help to understand step by step to configure the same.

That’s a complex task , but the cluster has a web interface for people with less experience.

  1. You need the high availability extension /in SLES 12 it was ‘ha_sles’ pattern/
  2. You have to setup the cluster . Use the ‘ha-cluster-init’ and pass everything necessary. As this is a VMware environment , you have to decide which approach will be taken:
    A) Do not use VmWare snapshots & use SBD for fencing, use btrfs snapshots
    B) Use Vmware snapahots (we use it for backup) & Do NOT use SBD for fencing (unless iSCSI is used which requires another machine), fencing mechanism will be Vmware based.

Depending on the above , ‘ha-cluster-init’ should be used with different options.
Setup /etc/hosts & Time sync (either chrony or ntp) - as these are critical for your cluster

  1. Define what the SAP App needs. I have only clusterised HANA , so I have no details about that.
    Usually a shared storage, Floating IP (cluster controlled), and start/stop script for the APP. The cluster supports systemd units, but they should be disabled on all nodes.

For shared storage, keep in mind that there are 2 approaches - HA-LVM (supported with 2 corosync rings) or DLM/CLVMd (no support for 2 corosync rings).

You can use the web interface (ha-cluster-init will tell you the IP + Port), bit channge the password for hacluster user on both nodees.

P.S.: You can use 3 VMs - 2 to host the app and the 3-rd one to have an odd number of cluster nodes - for best quorum calculation.

[QUOTE=strahil-nikolov-dxc;58334]That’s a complex task , but the cluster has a web interface for people with less experience.

  1. You need the high availability extension /in SLES 12 it was ‘ha_sles’ pattern/
  2. You have to setup the cluster . Use the ‘ha-cluster-init’ and pass everything necessary. As this is a VMware environment , you have to decide which approach will be taken:
    A) Do not use VmWare snapshots & use SBD for fencing, use btrfs snapshots
    B) Use Vmware snapahots (we use it for backup) & Do NOT use SBD for fencing (unless iSCSI is used which requires another machine), fencing mechanism will be Vmware based.

Depending on the above , ‘ha-cluster-init’ should be used with different options.
Setup /etc/hosts & Time sync (either chrony or ntp) - as these are critical for your cluster

  1. Define what the SAP App needs. I have only clusterised HANA , so I have no details about that.
    Usually a shared storage, Floating IP (cluster controlled), and start/stop script for the APP. The cluster supports systemd units, but they should be disabled on all nodes.

For shared storage, keep in mind that there are 2 approaches - HA-LVM (supported with 2 corosync rings) or DLM/CLVMd (no support for 2 corosync rings).

You can use the web interface (ha-cluster-init will tell you the IP + Port), bit channge the password for hacluster user on both nodees.

P.S.: You can use 3 VMs - 2 to host the app and the 3-rd one to have an odd number of cluster nodes - for best quorum calculation.[/QUOTE]

It seems that you can still use SBD for fencing mechanism, BUT you should add the shared disk as “Independant - Persistant”. This is also valid for all other shared disks -that’s way the OS can have Vmware-based snapshots (OS backup for example), while backup of the APP (which is on the shared storage) should be done by other means.