2222809 – SAP HANA database crashed immediately after a successful recovery and master index server failed to start again


Just after a successful database recovery the master index server crashed with the stack back trace from below. The index server will automatically be restarted several times but always fail for the same reason. Finally the hdbdaemon will stop starting the index server again and the system will never reach an operational status so that it can be used by business applications.


Stacktrace of crash: (2015-09-16 17:11:16 927 Local) —-> Pending exceptions (possible root cause) <—- exception 1: no.1000000 (Backup/Common/impl/BackupCom_Serializable.cpp:500) bool out of range: 235); $condition$=value <= 1 exception throw location:

1: 0x00007fb8477221ad in Backup::BackupCom_Serializable::GetValue(bool&)+0x49 at BackupCom_Serializable.cpp:500 (libhdbbackup.so)

2: 0x00007fb8476ed63a in Backup::BackupCatalog_LogMissingEntry::BackupCatalog_LogMissingEntry(Backup::BackupCom_Serializable const&)+0x126 at BackupCatalog_LogMissingEntry.cpp:146 (libhdbbackup.so)

3: 0x00007fb8475514d9 in Backup::BackupCom_StatisticsViewContent::deserialize(void const*, unsigned long, ltt::allocator&)+0xc15 at BackupCom_StatisticsViewContent.cpp:201 (libhdbbackup.so)

4: 0x00007fb8475c1223 in Backup::BackupMonitor_RequestHandler::handleRequest(TrexNet::Request&)+0x320 at BackupMonitor_RequestHandler.cpp:81 (libhdbbackup.so)


Other Terms

Recovery, backup catalog migration, log missing and scratch transfer


Reason and Prerequisites

  1. You are performing a database recovery
  2. You are using a revision of HANA SPS 10
  3. The affected database was upgraded from an SPS older than SPS 10
  4. The backup catalog to be used contains at least one entry of the type “log missing” created before SPS 10

The error occurred right after the recovery during transferring the backup catalog entries from the name server to the master index server for the purpose of showing them in the backup and recovery system views. The error is caused by the wrong interpretation of a migrated backup catalog entry of the type “log missing” which is rarely used in HANA.



A solution is available with SAP HANA revision 102.1 The bug fix also applies to customers that are already successfully migrated to HANA SPS 10 without using a database recovery.


If your system is still in operational state ONLINE you can check if you are potentially affected by this bug using the query statement from below. Keep in mind that this statement runs against an internal system view that needs to be fully qualified and ends with a “_”.


If at least one result entry is found you can truncate your backup catalog according to the description in the section “Housekeeping for Backup Catalog and Backup Storage” of the SAP HANA Administration Guide to get rid of these entries.

If you are facing the error scenario during a recovery and you are using file system based backups you could recreate the backup catalog according to the SAP note 1812057. Afterwards the recovery needs to be repeated with that new backup catalog.

If you are facing the error scenario during a recovery and you are using backint-based backups, you should open a SAP incident and request to forward this immediately to the HANA Development Support Organization.

After solving the issue using one of the workarounds it’s strongly recommended to create a new full data backup.


Leave a Reply