failback fails in clustered HANA setup

[Former Member]

Hi,

We have two HANA SPS12 servers configured in a cluster using the SLES 11 SP3 HA Extension.

Failing from node 1 to 2 works as expected, but failing back from node 2 to 1 fails with messages like "read phase '2', from takeover file" in the nameserver trace file.

It looks like the hdbnsutil -sr_takeover command does not fully complete leaving the database in an inaccessible state.

Anybody encountered this before?

Sander

[former_member205280]

Hi Sander,

Are you using the SAPHanaSR cluster agent package?

I'm using it in SLES 12 SP1, works well when using the AUTOMATED_REGISTER="true" option.

The system automatically takeover and re-register the primary/secondary system, works in both way from A - B or B - A. Which is very nice, but they only support single-box system installation right now.

Here're some white paper

https://www.suse.com/communities/blog/current-information-saphanasr-resource-agents-sap-hana-system-...

https://www.hds.com/en-us/pdf/best-practices/sap-hana-scale-up-high-availability-with-hana-system-re...

Regards,

Michael