Skip to main content

I have a commvault client that appears to be backing up the HANA database via the filesystem subclient, and I set the nFilterApplicationFiles setting and a full ran over the weekend and there was no change in the size and files backed up, and looking at the content of the backup, it appears there’s > 2 TB of data backed up in the fileystem subclient that’s in the /hana/shared/sdb] path. I was under the impression that nFilterApplicationFiles would prevent that… but the “database discovered” is located in the pseudoclient, not the SAP for HANA agent associated to the server level commvault client….

 

Questions:

  1. Does the nFilterApplicationFiles setting “work like it should” by not backing up the HANA database when the HANA pseudoclient is used as the backup method, and backups are “pushed” to commvault via the SAP/HANA side (using backint?).
  2. Are there other settings/configs to make it work?  Like something I would have to do in the “SAP for HANA” subclient? I coudl manually add in filesystem subclient level “ignore filters” of the DB paths, but I was hoping nFilterApplicationFiles would handle it dynamically

Hi @tigger2 

This key should exclude the paths under /hana/**/mnt00001 folders from FS backup on the HANA server even the database gets added under the pseudo client.

But for this we need at least one backup run from Commvault GUI, since that’s when we capture the topology of the HANA Database.

Which service pack you’re at on the SAP HANA client?

 

Thanks,

Sunil


For: “But for this we need at least one backup run from Commvault GUI”, the first (and all) backups are pushed from SAP/HANA side. Backups of the filesystem and HANA database have been happening for a long time (years), and I recently found and added the nFilterApplicationFiles setting to the commvault toplevel client.  However, the HANA team is refreshing the HANA environment so we are setting up new servers and pseudoclients, and I’d like to get them all set up “the same” and as properly as possible.

For: “Which service pack you’re at on the SAP HANA client?” We’re on v11 SP24.56

 


Ok.. that explains. Are these SAP HANA instances multi-tenant or single DB? If multi-tenant, are the SYSTEMDB and Tenant DBs discovered and shown in Commvault GUI hierarchy?

 

Thanks,

Sunil


For the one I specifically am asking about (or: that got me to post this question) is single tenant that has not been updated yet on the HANA side and I decided to test this setting by adding it to see how it would affect the backups (and there was no effect).

For the “new” ones we are setting up (cutting over for HANA upgrades), some are multi-tenant, some are single tenant (I think), but we definitely have multi-tenant ones in the “new” setups.


Okay. For this instance you’re testing, just open the instance properties in Commvault GUI and ‘Ok’ it.

This will trigger a validation on the client side which also collects the topology information. The FS backups further should exclude HANA folders.

 

Thanks,

Sunil


Opening the instance properties in Commvault GUI under the pseudoclient, and then hitting “ok” causes the Java UI to hang (hosted on the commserve) and the properties window to not respond nor close.  I had to manually kill the Java UI process.

But: from what you’re saying: the nFilterApplicationFiles should work, as long as a backup of the HANA database is done (via pseudoclient backup, pushed from HANA side) first, and then the Filesystem client should “see” it.

I believe the settings in the instance are not “all set” (like logins and passwords) because when SAP pushes the data to commvault, these are not needed (because commvault is not logging in to pull a backup) but that means that the instance properties cannot all validate?  The only things that should not be ‘set properly” are logins and passwords (at this level).  When we set up the “new” HANA pseudoclients when we typed in the instance name into the pseudocluent, it filled in everything else for us on teh instance properties.

 


nFilterApplicationFiles will work only when the Commvault instance is configured with valid login and password (or HDB Store key). Commvault validates the credentials passed, runs HANA query to find out the topology and then only we know what paths to exclude.

 

Thanks,

Sunil


ok… for the “

For this instance you’re testing, just open the instance properties in Commvault GUI and ‘Ok’ it.

This will trigger a validation on the client side which also collects the topology information. The FS backups further should exclude HANA folders.”

Thanks! We found that we were able to get 1 of these to work properly.  So it appears the permissions/setup on the HANA side for the account we are using for the instance setup is now working and one we are also using to configure a second instance is still not working, even though it's the same account name.

knowing it re-validates on hitting “ok” is perfect, so we can check the account works.  it's *really* annoying that when it doesn’t work, it takes the commvault Java UI like 3 minutes to time out to show an error, which just says ‘cannot connect” and doesn’t say why (well “cannot connect to DB, or sometimes “cannot connect to cilent”), or in a clean way that’s understandable or what it's trying to do to connect that is failing ( like “filed to execute /opt/abc def -ghi” )


Good to know you’re able to get your instance working.

Thanks for the feedback on the error propagation part. We will work on improving the error message to give more granular detail.

 

Thanks,

Sunil


Reply