cancel
Showing results for 
Search instead for 
Did you mean: 

Recovery Error

former_member183087
Participant
0 Kudos

Hi Experts

We have a situation to recover the HANA database but ending up below error, says Could not start master name server

We are using Rev 62 HANA One.

Our team was loading data into the system due to which the log areas were full and 100% utilized and the system crashed. During the crash the filesystem details are as below:

Filesystem        Size  Used Avail Use% Mounted on
/dev/hda1         9.9G  4.1G  5.3G  44% /
devtmpfs           30G  144K   30G   1% /dev
tmpfs              30G 0   30G   0% /dev/shm
/dev/md0          119G  113G 0 100% /hanalog
/dev/md1          197G   12G  176G   7% /hanadata
/dev/md2          197G  187G 0 100% /backup
/dev/xvdj          60G   22G   35G  38% /usr/sap

Tried to move some old log files to different location and tried stopping and starting which is failed to start. Now when we are trying to recover the database from the latest backup available we are getting the above error.

Can someone advice what is the issue and why we are not able to recover the system ? Is there is any workaround for the same ?

Last error log for recovery failure:

2013-10-07T04:56:49+00:00  P22344      14191465c0c INFO    RECOVERY command: RECOVER DATA USING FILE ('/backup/HDB/data/COMPLETE_DATA_BACKUP') CLEAR LOG
2013-10-07T04:56:49+00:00  P22344      14191465c0c INFO    RECOVERY state of service: nameserver, hanaserver:30001, volume: 0, RecoveryExecuteCatalogRecoveryInProgress
2013-10-07T04:56:49+00:00  P22344      14191465c0c INFO    RECOVERY state of service: nameserver, hanaserver:30001, volume: 0, RecoveryError
2013-10-07T04:56:49+00:00  P22344      14191465c0c ERROR   RECOVERY RECOVER DATA finished with error: [449] recovery strategy could not be determined, [110013] Unexpected end of backup

Regards

Srikar

Accepted Solutions (0)

Answers (1)

Answers (1)

swapan_saha
Advisor
Advisor
0 Kudos

Hi Srikar,

I see you both log volume (/hanalog) and backup volume (/backup - used for both data and log) are 100% full. I have a few comments and one suggestion here.

1. If you load data for the first time, you have  probably log enabled. Is there any reason not to disable logging for initial logging? It will save log and backup log volumes.

2. While the instance is down, you may stop the instance to stop billing while you are not using.

3. Please keep HANA One management console dashboard open so that you can monitor disk utilization too.

4. In the meantime, if you have a data backup, you may start a new HANA One instance and recover the backup data into a new instance using the blog I wrote:

Thanks,

Swapan

former_member183087
Participant
0 Kudos

Hi Swapan

Thanks, I want to understand the usage of /hanalog and /backup/HDB/log folders. I presume that when the log gets generated in the system it first gets written in folder /hanalog/HDB/log/mnt00001/hdb00001 (there are 4 folders in the folder mnt00001) and then gets backed up to folder /backup/HDB/log. Let me know if this the strategy HANA one is following.

There is no reason for us in enabling the logs during initial dataload. Unknowingly we did the load without disabling the logs.

Regards

Srikar

swapan_saha
Advisor
Advisor
0 Kudos

Hi Srikar,

You may look into Storage configuration section of Even though we stopped supporting PIOPS for log volume, the basic information is still relevant for the latest version of HANA One.

HANA One has log volume defined as /hanalog/HDB. Log backup is configured as automatic job for 15 mins. So, when HANA engine requires, log will be backed up at /backup/HDB/log. As you were generating logs in /hanalog/HDB, it was backing up at /backup/HDB/log till it ran out of the space.

Thanks,

Swapan