Determining Required Recovery files in HANA database

[davidebruno]

Hello all.

I want to create a list of HANA log files that I need in order to restore the Hana database to a specified timestamp.

Scenario

HANA 1.0 rev 74

Host: scale out x 6nodes

O.S.: SLES

Timestamp to restore the database: 2014-11-30 23:30:00.000

Last backup start at Nov 30 19:00

and finish at Nov 30 23:06

Following the SAP Note 1821207 (Determining Required Recovery files for HANA) I should use the command

hdbbackupdiag -f -d /usr/sap/SID/backup/log -u "2014-11-30 23:30:00"

but I have a strange result from this command!

I obtained the output attached as backupdiag.txt

the strange is that if I analyze the list (1270 files...), I see that for example the last file is created at

ls -lrt /usr/sap/SID/HDB20/backup/log/log_backup_8_0_129864262528_129881037696.1417450717272

Dec  1 17:18 /usr/sap/SID/HDB20/backup/log/log_backup_8_0_129864262528_129881037696.1417450717272

How is it possible that the creation date of the last log file that I need to recover in order to recover the database is greater than Nov 30 23:30 ??

I think there is some trouble..   🙂

I created a simple query on HDBSQL in order to double check the list of hana log to recover starting from the time of the start of the last backup.. and the result is different (query.txt) (the last file is created at Nov 30 23:09)

Is there anybody that can tell me the reason of this problem? I'm doing wrong to run the command hdbbackupdiag ?

Thank you very much

Davide

query.txt.zip

backupdiag.txt.zip

[Former Member · Accepted solution]

Dear Davide,

I think the issue comes from the fact that in your command line you did not specify from which catalog the tool reads the informations required.

The command line as it is in your case, let the hdbbackupdiag use the latest available backup catalog, which - obviously - contains also the information related to the log backups beyond the point-in-time you specified.

So, in order to have only the partial information you require, you need to modify the command line adding the "-c" option, specifying the catalog to read from.

For example, the command line would look like:

hdbbackupdiag -f -d /backup/<SID>/log -c log_backup_0_0_0_0.1418209887536 -u "2014-12-10 12:00:00"

The file "log_backup_0_0_0_0.1418209887536", is nothing but the first backup of the name server ( which also contains also the backup catalog ) executed beyond the desired P-I-T, that you can pick up from the catalog ( either with a query on the db or directly from SAP HANA Studio ).

If you take a look at the two files I've uploaded, you're gonna see the differences in terms of command line and in terms of results of the output, given that the desired P-I-T is still the same.

I hope this helps you, though I think this is - however - a bug of the executable, since the application should show me only the file interested by the "-u" switch regardless the backup catalog I'm using ( the latest one, contains also the old backup, so... ).

Please give it a try and let me know.

Regards,

Luciano

hdbbackupdiag_without_catalog.txt.zip

hdbbackupdiag_with_catalog.txt.zip

[davidebruno · Accepted solution]

You are the men! Thank you very very much!

I thought it was a bug in the handling of the catalog in the command,

but even with a message in OSS SAP I had solved

Now with you your help, I can determine the exact list of logs

thanks again

Davide BRUNO

[davidebruno]

This message was moderated.

[davidebruno]

If there is anyone that can help me, pleae could you try to replicate my command and tell me if the output is ok?

[lbreddemann]

Looks wrong indeed.

I'd recommend to check with a more current revision or to open a support incident.

- Lars

[davidebruno]

Thank you but we have also a SAP Hana rev. 85 (in Scale up scenario single node) with the same behaviour..