Skip to content

Commit efe4c43

Browse files
msftrobiromsftrobiro
committed
HANA multi-tier DR
1 parent 01c6347 commit efe4c43

File tree

2 files changed

+176
-0
lines changed

2 files changed

+176
-0
lines changed

‎articles/sap/workloads/disaster-recovery-sap-hana-suse.md

Lines changed: 176 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,176 @@
1+
---
2+
title: HSR third site with Pacemaker
3+
description: Extending highly-available SAP HANA solution with third site on SUSE systems.
4+
author: msftrobiro
5+
ms.author: robiro
6+
ms.service: sap-on-azure
7+
ms.subservice: sap-vm-workload
8+
ms.topic: how-to
9+
ms.date: 02/21/2023
10+
ms.custom: template-how-to-pattern
11+
---
12+
13+
# HSR third site with Pacemaker
14+
15+
This article describes requirements and setup of a third HANA replication site complimenting an existing SUSE Linux Enterprise Server (SLES) Pacemaker cluster.
16+
17+
## Overview
18+
19+
SAP HANA supports system replication with more than two sites connected. A third site can be added to an existing HSR pair, which is managed by Pacemaker in a highly-available setup. The third site can be deployed in a second Azure region for disaster recovery (DR) purposes.
20+
21+
Pacemaker and HANA cluster resource agent manages the first two sites. Any third, DR site, is not controlled by the cluster.
22+
23+
SAP HANA supports a third system replication site in two modes.
24+
- [Multi-target](https://help.sap.com/docs/SAP_HANA_PLATFORM/6b94445c94ae495c83a19646e7c3fd56/ba457510958241889a459e606bbcf3d3.html) replicates data changes from primary to more than one target system. Third site connected to primary, star form replication.
25+
- [Multi-tier](https://help.sap.com/docs/SAP_HANA_PLATFORM/6b94445c94ae495c83a19646e7c3fd56/f730f308fede4040bcb5ccea6751e74d.html) is a 2-tier system replication. A cascading, or sometimes referred to as chained setup, of 3 different HANA tiers. Third site connected to secondary.
26+
See [SAP HANA availability across Azure regions](./sap-hana-availability-across-regions.md#combine-availability-within-one-region-and-across-regions) for more details about HANA HSR within one region and across Azure regions.
27+
28+
## Prerequisites
29+
30+
Requirements for a third HSR site are different for HANA scale-up (single VM for each HANA system) and HANA scale-out (multiple VMs for each HANA system).
31+
32+
> [!NOTE]
33+
> Dependencies are valid for a Pacemaker enabled landscape. Without Pacemaker, only SAP's HANA version requirements apply.
34+
35+
- SAP HANA SPS 04 or newer is needed to use multi-target HSR with a Pacemaker cluster
36+
- Maximum one additional SAP HANA system replication connected from outside the Linux cluster
37+
- HANA scale-out only: SLES 15 SP1 or higher
38+
- HANA scale-out only: OS package SAPHanaSR-ScaleOut version 0.180 or higher
39+
- HANA scale-out only: SAP HANA HA provider SAPHanaSrMultiTarget in use. HANA HA provider SAPHanaSR is not multi-target aware for scale-out.
40+
41+
Pacemaker and HANA cluster resource agent manages only two sites. Third site is not controlled by Pacemaker cluster.
42+
43+
## HANA scale-up: Setup HANA multi-target system replication for DR purposes
44+
45+
With SAP HANA HA provider [SAPHanaSR](./sap-hana-high-availability.md#implement-hana-hooks-saphanasr-and-suschksrv), you can configure a third node for disaster recovery (DR) purposes. The Pacemaker environment is made aware of a HANA multi-target DR setup.
46+
47+
Failure of the third node won't trigger any cluster action. Cluster detects the replication status of connected sites and the monitored attribute for third site can change between SOK and SFAIL state. Any takeover tests to third/DR site or executing your DR process should first place the cluster resources into maintenance mode to prevent any undesired cluster action.
48+
49+
Example of a multi-target system replication system. For further information, see [SAP documentation](https://help.sap.com/docs/SAP_HANA_PLATFORM/4e9b18c116aa42fc84c7dbfd02111aba/2e6c71ab55f147e19b832565311a8e4e.html).
50+
![Example of a multi-target system replication system](./media/sap-hana-availability-two-region/saphanaavailability_hana_system_2region_ha_and_dr_multitarget_3vms.png)
51+
52+
1. Deploy Azure resources for the third node. Depending on your requirements, a different Azure region is often used for disaster recovery purposes.
53+
Steps required for the HANA scale-out on third site are mirroring steps to deploy the [HANA scale-up cluster](./sap-hana-high-availability.md#deploy-for-linux). Deploy the third node following the Azure infrastructure, operating system and HANA installation steps for first node of the Pacemaker cluster, with the following exceptions:
54+
- No load balancer deployed for third site and no integration with existing cluster load balancer for the VM of third site
55+
- OS packages SAPHanaSR, SAPHanaSR-doc and OS package pattern ha_sles aren't installed on third site VM
56+
- No integration into the cluster for VM or HANA resources of the third site
57+
- No HANA HA hook setup for third site in global.ini
58+
59+
2. Install SAP HANA on third node.
60+
Same HANA SID and HANA installation number must be used for third site.
61+
62+
3. With SAP HANA on third site installed and running, register the third site with the primary site.
63+
The example uses SITE-DR as the name for third site.
64+
```bash
65+
# Execute on the third site
66+
su - hn1adm
67+
# Make sure HANA is not running on the third site. If it is started, stop HANA
68+
sapcontrol -nr 03 -function StopSystem
69+
sapcontrol -nr 03 -function WaitforStopped 600 10
70+
# Register the HANA third site to the primary
71+
hdbnsutil -sr_register --name=SITE-DR --remoteHost=hn1-db-0 --remoteInstance=03 --replicationMode=async
72+
```
73+
74+
4. Verify HANA system replication shows both secondary and third site.
75+
```bash
76+
# Verify HANA HSR is in sync, execute on primary
77+
sudo su - hn1adm -c "python /usr/sap/HN1/HDB03/exe/python_support/systemReplicationStatus.py"
78+
# Third site, SITE-DR, will show up next to existing replication to SITE2 used by Pacemaker cluster.
79+
```
80+
81+
5. Check the SAPHanaSR attribute for third site. SITE-DR should show up with status SOK in the sites section.
82+
```bash
83+
# Check SAPHanaSR attribute on any cluster managed host (first or second site)
84+
sudo SAPHanaSR-showAttr
85+
# Expected result
86+
# Global cib-time maintenance
87+
# --------------------------------------------
88+
# global Tue Feb 21 19:28:21 2023 false
89+
#
90+
# Sites srHook
91+
# -----------------
92+
# HN1-SITE1 PRIM
93+
# HN1-SITE2 SOK
94+
# SITE-DR SOK
95+
```
96+
97+
Cluster detects the replication status of connected sites and the monitored attributed can change between SOK and SFAIL. No cluster action if the replication to DR site is impacted.
98+
99+
### Auto-registering third site
100+
101+
During planned or unplanned takeover event between the two Pacemaker cluster sites, HSR to third site will be also interrupted. For Pacemaker setting `AUTOMATED_REGISTER = yes|no` can control registration between Pacemaker controlled sites 1 and 2. The parameter does not control HSR replication to the cluster controlled third site.
102+
103+
SAP provides since HANA 2 SPS 04 parameter `register_secondaries_on_takeover`. With the parameter set to value `true`, after HSR takeover between cluster sites 1 and 2, HANA will register the third site on the new primary automatically to keep a HSR multi-target setup. Configure HANA parameter `register_secondaries_on_takeover = true` configured in `[system_replication]` block of global.ini on both SAP HANA sites in the Linux cluster. Both first and second node/site need the parameter in the respective HANA global.ini configuration file.
104+
105+
For HSR multi-tier, no automatic registration exist for the third site. Manual registration of third site to the current secondary, to keep HSR replication chain, is needed.
106+
107+
## HANA scale-out: Setup HANA multi-target system replication for DR purposes
108+
109+
With SAP HANA HA provider [SAPHanaSrMultiTarget](./sap-hana-high-availability-scale-out-hsr-suse.md#implement-hana-ha-hooks-saphanasrmultitarget-and-suschksrv), a third HANA scale-out site can be added. This third site is commonly used for disaster recovery (DR) in another Azure region. The Pacemaker environment is made aware of a HANA multi-target DR setup.
110+
111+
Failure of the third site won't trigger any cluster action. Cluster detects the replication status of connected sites and the monitored attribute for third site can change between SOK and SFAIL state. Any takeover tests to third/DR site or executing your DR process should first place the cluster resources into maintenance mode to prevent any undesired cluster action.
112+
113+
Example of a multi-target system replication system. For further information, see [SAP documentation](https://help.sap.com/docs/SAP_HANA_PLATFORM/4e9b18c116aa42fc84c7dbfd02111aba/2e6c71ab55f147e19b832565311a8e4e.html).
114+
![Example of a multi-target system replication system](./media/sap-hana-high-availability/sap-hana-high-availability-scale-out-hsr-suse-multi-target.png)
115+
116+
1. Deploy Azure resources for the third site. Depending on your requirements, a different Azure region is often used for disaster recovery purposes.
117+
Steps required for the HANA scale-out on third site are mirroring steps to deploy the [HANA scale-out cluster](./sap-hana-high-availability-scale-out-hsr-suse.md#set-up-the-infrastructure). Deploy the third site following the Azure infrastructure, operating system and HANA installation steps for SITE1 of the scale-out cluster, with the following exceptions:
118+
- No load balancer deployed for third site and no integration with existing cluster load balancer for the VMs of third site
119+
- OS packages SAPHanaSR-ScaleOut, SAPHanaSR-ScaleOut-doc and OS package pattern ha_sles aren't installed on third site VMs
120+
- No majority maker VM for third site, as there's no cluster integration
121+
- NFS volume /hana/shared for third site exclusive use must be created
122+
- No integration into the cluster for VMs or HANA resources of the third site
123+
- No HANA HA hook setup for third site in global.ini
124+
125+
Same HANA SID and HANA installation number must be used for third site.
126+
127+
2. With SAP HANA scale-out on third site installed and running, register the third site with the primary site.
128+
The example uses SITE-DR as the name for third site.
129+
```bash
130+
# Execute on the third site
131+
su - hn1adm
132+
# Make sure HANA is not running on the third site. If it is started, stop HANA
133+
sapcontrol -nr 03 -function StopSystem
134+
sapcontrol -nr 03 -function WaitforStopped 600 10
135+
# Register the HANA third site to the primary
136+
hdbnsutil -sr_register --name=SITE-DR --remoteHost=hana-s1-db1 --remoteInstance=03 --replicationMode=async
137+
```
138+
139+
3. Verify HANA system replication shows both secondary and third site.
140+
```bash
141+
# Verify HANA HSR is in sync, execute on primary
142+
sudo su - hn1adm -c "python /usr/sap/HN1/HDB03/exe/python_support/systemReplicationStatus.py"
143+
# Third site, SITE-DR, will show up next to existing replication to SITE2 used by Pacemaker cluster.
144+
```
145+
146+
4. Check the SAPHanaSR attribute for third site. SITE-DR should show up with status SOK in the sites section.
147+
```bash
148+
# Check SAPHanaSR attribute on any cluster managed host (first or second site)
149+
sudo SAPHanaSR-showAttr
150+
# Expected result
151+
# Global cib-time maintenance prim sec sync_state upd
152+
# ---------------------------------------------------------------------
153+
# HN1 Fri Jan 27 10:38:46 2023 false HANA_S1 - SOK ok
154+
#
155+
# Sites lpt lss mns srHook srr
156+
# ------------------------------------------------
157+
# SITE-DR SOK
158+
# HANA_S1 1674815869 4 hana-s1-db1 PRIM P
159+
# HANA_S2 30 4 hana-s2-db1 SWAIT S
160+
```
161+
162+
Cluster detects the replication status of connected sites and the monitored attributed can change between SOK and SFAIL. No cluster action if the replication to DR site is impacted.
163+
164+
### Auto-registering third site
165+
166+
During planned or unplanned takeover event between the two Pacemaker cluster sites, HSR to third site will be also interrupted. For Pacemaker setting `AUTOMATED_REGISTER = yes|no` can control registration between Pacemaker controlled sites 1 and 2. The parameter does not control HSR replication to the cluster controlled third site.
167+
168+
SAP provides since HANA 2 SPS 04 parameter `register_secondaries_on_takeover`. With the parameter set to value `true`, after HSR takeover between cluster sites 1 and 2, HANA will register the third site on the new primary automatically to keep a HSR multi-target setup. Configure HANA parameter `register_secondaries_on_takeover = true` configured in `[system_replication]` block of global.ini on both SAP HANA sites in the Linux cluster. Both SITE1 and SITE2 need the parameter in the respective HANA global.ini configuration file.
169+
170+
For HSR multi-tier, no automatic registration exist for the third site. Manual registration of third site to the current secondary, to keep HSR replication chain, is needed.
171+
172+
## Next steps
173+
174+
- [Disaster recovery overview and infrastructure](./disaster-recovery-overview-guide.md)
175+
- [Disaster recovery for SAP workloads](./disaster-recovery-sap-guide.md)
176+
- [High-availability architecture and scenarios for SAP NetWeaver](./sap-hana-availability-across-regions.md)

‎articles/sap/workloads/media/sap-hana-high-availability/sap-hana-high-availability-scale-out-hsr-suse-multi-target.png

30 KB
Loading

0 commit comments

Comments
 (0)