About the participating ASE in HADR

[DilipVoora]

Hi Experts,

I configured HADR(Sync) in my test machines and I would like to get the below questions clarified from you.

Details:

A) Cluster ID database : LE1
B) Host 1: Test1.dba.com
Primary dataserver : ASE1 (16.0 SP03 PL05 with 4K page size)
Primary site name : Andhra

C) Host2: Test2.dba.com
Target/Companion dataserver : ASE2 (16.0 SP03 PL05 with 4K page size)
Target/Companion site name : NewDelhi

Questions:

1. While checking rs_databases&rs_helpdb (output attached) I see two extra dataservers (LE1_NewDelhi_R1 & LE1_Andhra_R2) and I would like to know from where these dataservers came from and their exact use?
2. These two dataservers are using the same port number of the dataserver that are runing on the hosts. Interfaces file attached.
3. I also see the database replication definitions corresponding to these dataservers when checked from rs_helpdbrep and they are using same replication server for PRS & RRS and as per the architecture data is getting replicating to the database in comapnion ASE through the replciation definition and subscription realted to these extra dataservers.

I tried but I couldn't understand. So, requesting to get advised.

Regards,
Dilip Voora

[c_baker]

I am answering based on ASE 16.0 SP04PL04.

The HADR components refer to each other using logical names maintained in a $SYBASE/DM/interfaces file and do not use the $SYBASE/interfaces file when defining the connectivity for any component. This also keeps the $SYBASE/interfaces file clear of HADR definitions.

As you can see, the $SYBASE/DM/interfaces logical names for the ASEs are constructed from the cluster, site and route names.

The ASE instances themselves are normally started using the 'master' entries in the $SYBASE/interfaces files. Other HADR components may use the 'master' entries in the $SYBASE/DM/interfaces files when starting, but connections to the ASEs (and other HADR components) will use the 'query' values defined in the $SYBASE/DM/interfaces file.

Your 'extra' dataservers are simply internal references to the original dataservers, just using a separate interfaces file.

Chris

[DilipVoora]

Hi Chris,

I understood from your comment that HADR components refer to each other using logical names and just found the same in HADR user guide (sorry I didn't read it completely before putting this post).

If we consider the scenario (ASE1 is primary and ASE2 is standby) I see the data flow like the below from the active repserver running on the host2.

ASE1.D30(primary) --->LE1_Andhra_R2_D30.D30(active repserver on host2) ---->LE1_NewDelhi_D30.D30 ---->ASE2.D30(target)

My question here is, why LE1_NewDelhi_D30 is cosnidered for replicating the data though it is not logical ?

Regards,
Dilip Voora

[wajeeh_samdani]

Hi Dilip,

The path you described is not correct as "LE1_Andhra_R2_D30.D30" is not the repserver on host2 as we can see from the interfaces files that your repservers run on the port 5005 and your ASEs run on port 5000. These *R1 and *R2 are used for internal communications between ASE's for example for startup procedure to check Standby ASE availability to avoid split brain scenario.

The correct path is simply:

ASE1.D30(primary) ---->LE1_NewDelhi_D30.D30 ---->ASE2.D30(target)

Regards,

Wajeeh

[c_baker]

Are you seeing 'LE1_Andra_R2_D30' or 'LE1_NewDelhi_D30' dsnames that do not have $SYBASE/DM/interfaces entries (i.e. dsnames that appear to have the db appended to the logical server name)?

You are running ASE 16.0 SP03PL05 (and I assume the corresponding SRS version). This version is very old (over 5 years old) and many improvements/fixes have been made to HADR over that time.

Have you been able to reproduce the observation in a current release (current SP03 release is ASE 16.0 SP03PL14 - released June 2023)? Is this a Business Suite system?

Chris