IBM Support

IC84314: NUMA API incompatibility on SLES11 and RHEL6

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as requirement.

Error description

  • The NUMA APIs that DB2 uses to provide NUMA support have an
    undocumented dependency between the build platform used to
    create DB2 libraries, and the kernel used while running DB2.
    
    The kernels on SLES11 and RHEL6 violate this dependency and are
    thus incompatible with DB2's NUMA support.
    
    If you attempt to enable NUMA support (via the
    DB2_RESOURCE_POLICY registry variable) on either of these
    platforms, you will encounter either of the following errors
    during db2start or db2 activate database.
    
    Error #1:
    
    2012-01-06-02.25.55.633828-300 E14095E469          LEVEL: Severe
    (OS)
    PID    : 49981                TID  : 139760697861888PROC :
    db2sysc
    INSTANCE: db2inst1            NODE : 000
    FUNCTION: DB2 UDB, oper system services,
    sqloIsCpuSetContainedInNUMANode, probe:10
    MESSAGE : ZRC=0x83000022=-2097151966
    CALLED  : OS, -, numa_node_to_cpus                OSERR: ERANGE
    (34)
    DATA #1 : Hexdump, 8 bytes
    0x00007F1C92BF9790 : 0000 0000 0000 0000
    ........
    
    Error #2:
    
    2012-01-04-10.31.53.765427-360 E10417E1419        LEVEL: Severe
    PID    : 10685                TID  : 140228274677504PROC :
    db2sysc 0
    INSTANCE: db2inst1            NODE : 000          DB  : SAMPLE
    APPHDL  : 0-7                  APPID: *LOCAL.sample.120104163153
    
    AUTHID  : DB2INST1
    EDUID  : 20                  EDUNAME: db2agent (TPCC) 0
    FUNCTION: DB2 UDB, SQO Memory Management, sqlo_numa_error,
    probe:10
    DATA #1 : <preformatted>
    Error encountered during NUMA operation: 'mbind' errno = 22
    CALLSTCK: (Static functions may not be resolved correctly, as
    they are resolved to the nearest symbol)
      [0] 0x00007F8979F5F693 pdLogPrintf + 0x3B3
      [1] 0x00007F897BA46D00 numa_error + 0x34
      [2] 0x00007F897A4CB09D
    _ZN12SQLO_LIBNUMA22numa_interleave_memoryEPvmP10nodemask_t +
    0x13
      [3] 0x00007F897BA7D193
    /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0x2871193
      [4] 0x00007F897A4C9D28
    _Z25sqloGetSharedMemoryFromOsP17SMemSharedOptions + 0x1B4
      [5] 0x00007F897A4C9795
    /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0x12BD795
      [6] 0x00007F897A4C7322 sqlogmshr + 0x880
      [7] 0x00007F897A0C2A74
    /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0xEB6A74
      [8] 0x00007F897A0C0BC3
    _ZN16sqeLocalDatabase12FirstConnectEP8SQLE_BWARcP8sqeAgentP8sqlo
    _gmtii + 0x45B
      [9] 0x00007F897A0B8CF0
    _ZN8sqeDBMgr23StartUsingLocalDatabaseEP8SQLE_BWAP8sqeAgentRccP8s
    qlo_gmt + 0xC7C
    

Local fix

  • Disable NUMA support, by unsetting the DB2_RESOURCE_POLICY
    registry variable.
    

Problem summary

  • ****************************************************************
    * USERS AFFECTED:                                              *
    ****************************************************************
    * PROBLEM DESCRIPTION:                                         *
    * The NUMA APIs that DB2 uses to provide NUMA support have an  *
    * undocumented dependency between the build platform used to   *
    * create DB2 libraries, and the kernel used while running DB2. *
    *                                                              *
    * The kernels on SLES11 and RHEL6 violate this dependency and  *
    * are                                                          *
    * thus incompatible with DB2's NUMA support.                   *
    *                                                              *
    * If you attempt to enable NUMA support (via the               *
    * DB2_RESOURCE_POLICY registry variable) on either of these    *
    * platforms, you will encounter either of the following errors *
    * during db2start or db2 activate database.                    *
    *                                                              *
    * Error #1:                                                    *
    *                                                              *
    * 2012-01-06-02.25.55.633828-300 E14095E469          LEVEL:    *
    * Severe                                                       *
    * (OS)                                                         *
    * PID    : 49981                TID  : 139760697861888PROC :   *
    * db2sysc                                                      *
    * INSTANCE: db2inst1            NODE : 000                     *
    * FUNCTION: DB2 UDB, oper system services,                     *
    * sqloIsCpuSetContainedInNUMANode, probe:10                    *
    * MESSAGE : ZRC=0x83000022=-2097151966                         *
    * CALLED  : OS, -, numa_node_to_cpus                OSERR:     *
    * ERANGE                                                       *
    * (34)                                                         *
    * DATA #1 : Hexdump, 8 bytes                                   *
    * 0x00007F1C92BF9790 : 0000 0000 0000 0000                     *
    * ........                                                     *
    *                                                              *
    * Error #2:                                                    *
    *                                                              *
    * 2012-01-04-10.31.53.765427-360 E10417E1419        LEVEL:     *
    * Severe                                                       *
    * PID    : 10685                TID  : 140228274677504PROC :   *
    * db2sysc 0                                                    *
    * INSTANCE: db2inst1            NODE : 000          DB  :      *
    * SAMPLE                                                       *
    * APPHDL  : 0-7                  APPID:                        *
    * *LOCAL.sample.120104163153                                   *
    *                                                              *
    * AUTHID  : DB2INST1                                           *
    * EDUID  : 20                  EDUNAME: db2agent (TPCC) 0      *
    * FUNCTION: DB2 UDB, SQO Memory Management, sqlo_numa_error,   *
    * probe:10                                                     *
    * DATA #1 : <preformatted>                                     *
    * Error encountered during NUMA operation: 'mbind' errno = 22  *
    * CALLSTCK: (Static functions may not be resolved correctly,   *
    * as                                                           *
    * they are resolved to the nearest symbol)                     *
    *   [0] 0x00007F8979F5F693 pdLogPrintf + 0x3B3                 *
    *   [1] 0x00007F897BA46D00 numa_error + 0x34                   *
    *   [2] 0x00007F897A4CB09D                                     *
    * _ZN12SQLO_LIBNUMA22numa_interleave_memoryEPvmP10nodemask_t + *
    * 0x13                                                         *
    *   [3] 0x00007F897BA7D193                                     *
    * /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0x2871193         *
    *   [4] 0x00007F897A4C9D28                                     *
    * _Z25sqloGetSharedMemoryFromOsP17SMemSharedOptions + 0x1B4    *
    *   [5] 0x00007F897A4C9795                                     *
    * /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0x12BD795         *
    *   [6] 0x00007F897A4C7322 sqlogmshr + 0x880                   *
    *   [7] 0x00007F897A0C2A74                                     *
    * /home/db2inst1/sqllib/lib64/libdb2e.so.1 + 0xEB6A74          *
    *   [8] 0x00007F897A0C0BC3                                     *
    * _ZN16sqeLocalDatabase12FirstConnectEP8SQLE_BWARcP8sqeAgentP8 *
    * sqlo                                                         *
    * _gmtii + 0x45B                                               *
    *   [9] 0x00007F897A0B8CF0                                     *
    * _ZN8sqeDBMgr23StartUsingLocalDatabaseEP8SQLE_BWAP8sqeAgentRc *
    * cP8s                                                         *
    * qlo_gmt + 0xC7C                                              *
    ****************************************************************
    * RECOMMENDATION:                                              *
    * Discontinue the use of NUMA memory affinity support, but     *
    * unsetting the DB2_RESOURCE_POLICY registry variable.         *
    ****************************************************************
    

Problem conclusion

  • This incompatiblity will be resolved in future releases through
    a combination of DB2 code changes and the use of newer build
    platforms.
    

Temporary fix

Comments

  • Will be addressed by DB2 code changes (use of NUMA v2 APIs) and
    upgrades to DB2 build environments (to avoid kernel/runtime data
    structure mismatch.)
    

APAR Information

  • APAR number

    IC84314

  • Reported component name

    DB2 FOR LUW

  • Reported component ID

    DB2FORLUW

  • Reported release

    A10

  • Status

    CLOSED REQ

  • PE

    NoPE

  • HIPER

    NoHIPER

  • Special Attention

    NoSpecatt

  • Submitted date

    2012-06-13

  • Closed date

    2012-12-17

  • Last modified date

    2012-12-17

  • APAR is sysrouted FROM one or more of the following:

    IC80998

  • APAR is sysrouted TO one or more of the following:

Fix information

Applicable component levels

[{"Line of Business":{"code":"LOB10","label":"Data and AI"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SSEPGG","label":"DB2 for Linux- UNIX and Windows"},"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"10.1"}]

Document Information

Modified date:
19 September 2021