Pacemaker is an open source cluster resource manager software since 2004. IBM integrated Pacemaker with DB2 starting DB2 version 11.5.4.0. In version 11.5.4.0 Pacemaker was introduced as Technical Preview and then released for production use in 11.5.5.0. If you want to use open source Pacemaker cluster with DB2 you can use it but for Pacemaker cluster issues, you will not get support from IBM. Pacemaker can be used to automate failover in HADR env and setup mutual failover env.
Why to use DB2 Pacemaker:
1. Cloud compatible.
DB2 Pacemaker cluster is cloud compatible and you can use it in any private and public cloud.
2. No additional license or support cost.
Even though Pacemaker is an open source cluster software, if you plan to use Pacemaker provided by Linux distros, there might be additional license cost and support cost. With DB2 Pacemaker there is no additional license or suppose cost.
3. Better support from IBM
If you encounter any issue with DB2 Pacemaker, your ticket will be handled by DB2 support team and there is no dependency on non DB2 support team which should result in faster resolution.
4. Simpler Architecture
TSA/RSCT has multiple components serving difference purposes. Pacemaker architecture is a lot leaner and less complicated than TSA/RSCT.
5. Easier Problem Determination
TSA has multiple logs and traces files, some of them in binary format which needs to be formatted before starting with troubleshooting. These TSA log and trace files are available in different directory making it more complicated. In pacemaker all logs are present in one location and it is in text format with timestamp for each entry. Pacemaker log is present in path /var/log/pacemaker/pacemaker.log
6. Faster Recovery Time
Pacemaker is able to recover faster from failure in both manual takeover and automated failover due to system failure. Pacemaker recovery time is approximately 30% faster than TSA in lab env.
Pacemaker Process:
There are 7 processes to keep pacemaker cluster function on each node in the cluster i.e. pacemakerd, pacemaker-controld, pacemaker-schedulerd, pacemaker-based, pacemaker-attrd, pacemaker-execd, pacemaker-fenced.
- pacemakerd: It is the watchdog of all pacemaker process, it cleans up and respawns them if any of the process is missing.
- pacemaker-controld: It maintains cluster membership and orchestrate actions. One of the controld process in the cluster will be elected as Domain Leader (DL).
- pacemaker-schedulerd: It checks resource states against configured desired state and advise Domain leader to take action.
- pacemaker-based: Pacemaker resource state and config are maintained in memory as well as on disk. On disk it is written in cluster information base file (CIB.xml). It is located in /var/lib/pacemaker/cib/cib.xml on each node. This deamon reads/write from CIB.xml and ensure CIB is synchronized across the cluster.
- pacemaker-attrd: This deamon maintains resource states/configuration in memory. It can make request to update CIB if require.
- pacemaker-execd: It executes actions programmed by the application through resource agent. E.g When an instance needs to be started, pacemaker invokes DB2 instance resource agent to start it.
- pacemaker-fenced: It fences agent on local node.
Below limitations are applicable to DB2 version 11.5.8.0 or earlier. DB2 11.5.8.0 is the latest DB2 version while publishing this blogs.
- Supported OS: Currently Pacemaker only supports RHEL 8.1 and above and SLES 15 SP1 or above.
- Pacemaker cannot be used to setup HA for DPF env or in PureScale cluster.
- Mutual failover feature is unavailable in DB2 11.5.7.0 or earlier.
- Prior to DB2 11.5.6.0, separate binaries needs to be downloaded for DB2 Pacemaker installation.
Above limitation will get fixed in future release of DB2.
Check blog Configure DB2 Pacemaker for HADR for step by step instructions to configure DB2 Pacemaker for HADR.
If you liked this blog and interested in knowing more about DB2, please subscribe by clicking on Subscribe to ChoudharySumit.com by Email.
No comments:
Post a Comment