Backint preventing SAP HANA services from starting ( DB hangs )

[Former Member]

Hello,

We are currently using backint for our full and log backups with SAP HANA SPS 9. They both work fine. We started having issues when we stop and start the SAP HANA database. It hangs and the following services do not start:

daemon - yellow

xsengine - yellow

webdispatcher - red

Initially, we thought it was related to our interfaces ( eth1 - eth3 ). While trouble shooting, we downed these mentioned interfaces and it allowed SAP HANA to start w/o issues. But...we have other tiers ( sandbox, quality and development ) using SPS 9 where we do not encounter this issue. The only difference with those tiers, we do not utilize the HP DP backint. So, on a hunch, we kept the interfaces up ( eth1 - eth3 ) during a stop and restart with a server giving us issues but changed the following parameter before re-starting:

[backup]

log_backup_using_backint = false

The server came online normal w/o any issues. We stopped it again, changed the parameter back to ( log_backup_using_backint = true ), and the hanging issue returned ( daemon - yellow, xsengine - yellow and webdispatcher - red ). Having to change this parameter every time we stop and start an SAP HANA DB is not viable. It will create gaps in our transaction logs ( backint vs local ) since SAP HANA automatically backs up log segments during a restart.

Has anyone encountered this issue with backint ? Is there a work-around ?

Thank you in advance,

/gxo13

[Bojan-lv-85]

Hello Gregorio,

the HANA instance should not hang during startup after the parameter-change, we definetively would need to investigate on what exactly is going wrong there.

Please update this discussion with the incident number in case you go for a Support Incident.

Thank you!

BR, Bojan

[Former Member]

Thank you all for the feedback. We did not see this anomaly in our test server. We are no longer seeing it with our other tiers too. Whatever was impeding the backint process from communicating with HP Data Protector was inadvertently fixed recently. The problem is that we don't know what got fixed in the process ( maybe a network issue ).

[former_member183326]

I think the best people to comment on this would be your Backint Vendor.

This does look like a Backint issue. I would suggest opening a ticket with your Backint provided rather than SAP.

[Former Member]

Hi Gregorio,

We are using SAP HANA landscapes on Distributed Single container system and Scale up Multi Container Database(MDC), these systems are being backed up using BACKINT using an EMC tool

We have been on SPS 07,08, 09 and 10 now. I must say we have not experienced such an issue during startup on any of our systems

I have not seen any reference for such an issue documented on SAP side yet

The only issue I am aware of with the BACKINT configuration on SPS09 is the MVCC garbage collection getting blocked when backint data backups are in progress while the Auditing is enabled(refer note below)

http://service.sap.com/sap/support/notes/2122316

I would suggest you to open a SAP Support case and share your findings, they should be able to validate if this is indeed an issue with BACKINT of just with the vendor you are using currently

Sunil

[former_member182967]

Hello Gregorio,

The Redo Log is backed up during database start. For log_backup_using_backint = true, the log should be backed up to your backup server.

You can check the following parts (in case log_backup_using_backint = true)

1. The connection to your backup server is OK or not during database start.

2. Check the backint configuration is correct or not.

3. Check the backlog, if note 2233763 - [447] backup could not be completed, [110503] Backint cannot execute /usr/sap/<SID>/SYS/gl... helps or not.

Regards,

Ning