Not having much hands on hadr, I have few questions about it.

1. two computers communicate (send-recieve) data using port (55000) between them. is that port number reserved or can be changed. Does it mean that hadr does never use standard connection port (50000) ??

2. Once there is a problem , standby ( computer 2) takes over , it becomes primary and the one was primary ( computer 1) before becomes standby. In this situation old primary gets into rollforward ( being standby) and logs start flowing from computer 2 to computer 1. The takeover happened because of some issue (malfunction) on the database. let us say the server instance goes down ( crashes ). It seems that there will be hardly any takeover without an issue. that means computer 1 has to be fixed and brought up to standby role. Please correct me where I may be wrong.

3. what re-routs the applications to new primary ??

4. Is there a possibility of standby falling out of sync in any case ; e.g. transactions (users) increase ? What can be the work around ??

Please help clear the concept.

Regards

DBFinder

1. HADR has it own ports separate from the database instance port. Usually there are two ports defined for HADR, one for local and one remote. You can use any port numbers for the instance or HADR so long as it is not being used by another service on your server.
2. HADR itself does not initiate the Takeover command. You must have some kind of cluster manager software that detects a server problem and issues the HADR Takeover command. IBM provides TSA (Tivoli System Automation) software with DB2 at no extra charge for this purpose, or you can use some other cluster management software or scripts. Obviously, you also issue the HADR Takeover command manually.
3. There are two basic ways. One is to use ACR (Automatic Client Reroute), which can you read about in the IBM documentation. The other is use a virtual IP address that is associated with one of the servers at any one time, and when the cluster management software wants to do a HADR Takeover, it also moves the virtual IP address to the new primary server. It is also possible (and actually recommended) that both VIP and ACR be used at the same time to facilitate retry logic (in this case, the primary and alternate server names are the same, which is the VIP host name). Moving the actual server IP address or hostname with a VIP would require root access and you can discuss this with a system adminstrator.
4. It is possible for HADR to get out of synch if the standby is down for so long that logs are no longer available on primary to send to the standby. If this happens, you must reinitialize the HADR the same way it was created (backup of primary and restore to standby). The backup can be done online without an outage.

Thanks Marcus,

Very clear and precise !

Is there a way that I can test TSA and ACR ( items 2 and 3 above ) in my personal test lab ? I mean, if these can be installed and tested as free trial downloads ??

Thanks you your time again

Regards
DBFinder

TSA comes along with the DB2 server installation. ACR is a feature of the DB2 client. In other words, yes, you can test it and no additional installation is required.

Many thanks Nick,

I have not noticed TSA in a trial downloads upto data server 9.5 fixpack 5.

May be I need to double check , with custom options !!


Thanks anyways.


Regards
DBFinder

Installing later versions IBM Tivoli System Automation for Multiplatforms (SA MP) Base Component

Thank you

DBFinder