Skip to Content
avatar image
Former Member

TSM restore fails for recover from different HDB-system

We try to recover a fulldatabackup of a different database to a HDB via backint from TSM.

It's a scaleout-system with 8 Nodes (1 master 7 worker, no spare)

The Configuration File of the source-system is implemented (/usr/sap/SID/SYS/global/hdb/opt/hdbconfig/initSID.utl also the .bki, Password is set) dsmc-Connection to TSM works and the Backupcatalog could be read.

RECOVERY command: RECOVER DATA USING SOURCE 'SID' USING BACKUP_ID xxxxxxxxxxxxxxxx USING CATALOG BACKINT USING DATA PATH ('/usr/sap/SID/SYS/global/hdb/backint/') CLEAR LOG

The Recovery begins and is able to restore several small Services like nameserver, scriptserver, e.g.) some indexservices are calculating others are already restoring (5-10%)

Unfortunately after ~5 minutes the Recovery stops with:

RECOVERY RECOVER DATA finished with error: [448] recovery could not be completed, volume 10, reached log position 0, [110512] Backint reported 'BACKINT recovery job into /usr/sap/SID/SYS/global/hdb/backint/backup_SID_xxxxxx_databackup_10_1 failed with error code ERROR' in file '/var/tmp/hdbbackint_SID.JCOzeZ'

The tmp-error file could not be found afterwards.

The TSM-Pipes in the backint-directory are missing, manually creation and restart of the recovery has the same effect, pipes are gone recovery stops.

Any known issues like that? Are there parameters for the hana to get more tries or time to deal with the TSM-Data? Timeout-Parameter in the python recovery script was already set from 120 to 3600: no effect

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

2 Answers

  • Best Answer
    avatar image
    Former Member
    Mar 16 at 10:14 AM

    The direct TSM-Recovery was not successful, yet, but it was possible to restore all the single TSM-Backupfiles backup_SID_xxxxxxxx_databackup_x_1 to a NFS-Mount, by TSM Client.

    From there a Hana-Recovery from file was successfully recovered. It seems to be a problem for the target-worker-nodes to get the source-fulldatabackup-files from the TSM with tdp_hana.

    Add comment
    10|10000 characters needed characters exceeded

  • avatar image
    Former Member
    Mar 13 at 04:56 PM

    There are some other var/tmp files saying:

    BKI4009E: Not enough space to write File '/usr/sap/SID/SYS/global/hdb/backint/backup_SID_xxxxxxxx_databackup_3_1'. Possible reasons: disk full or ulimit exeeded.

    We can not find disk space shortage or disk usage changes due to recovery also the ulimits are set like default. Nodes are SLES 11.4 . Maybe ulimit pipe size (512 Bytes) 8 could be a hint, but is not changable [ulimit: pipe size: cannot modify limit: Invalid argument]
    Add comment
    10|10000 characters needed characters exceeded