IBM Support

PM14503: CLEARCASE MVFS ALLOCATES BETA SEMAPHORES INEFFICIENTLY FOR MNODE ALLOCATION ON HP-UX 11.31 IA SYSTEMS

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as program error.

Error description

  • ClearCase MVFS allocates Beta Semaphores inefficiently for mnode
    allocation on HP-UX 11.31 IA systems
    
    
    ClearCase 7.0.1.3
    
    
    HP-UX 11.31 IA
    
    
    Problem Explained:
    
    Inefficient use of Beta Semaphores, when MVFS allocates mnodes,
    causes the lock class information hash table to increase in size
    with each new, unique mnode allocation.  As ClearCase is used,
    say within a build environment, the lock class information hash
    table increases exponentially: Subsequently, searching the hash
    table for entries becomes prohibitive, and the system
    performance begins to suffer.
    
    Workaround:  NONE
    

Local fix

Problem summary

  • ****************************************************************
    * USERS AFFECTED:                                              *
    ****************************************************************
    * PROBLEM DESCRIPTION:                                         *
    ****************************************************************
    * RECOMMENDATION:                                              *
    ****************************************************************
    The issue is caused due to the way MVFS allocates locks
    (beta semaphores) to protect its internal data structures.
    While allocating the locks, MVFS associated every lock with
    a unique name. This was primarily to aid debugging of locks.
    On HP-UX 11.31 while initializing the beta semaphore, the
    kernel creates a data structure of its own using the
    parameters passed to the initialization routine. One of the
    parameters that is used is name of the lock. This data
    structure is maintained as a hash. No other platform
    (including HP-UX 11.23 or HP-UX 11.11) use the information
    passed to the initialization routine this way. Entries in
    the hash are flushed only when system is rebooted. With
    every lock allocation, the size of the hash increases, hence
    the time taken to allocate locks increases exponentially.
    

Problem conclusion

  • A fix is available in ClearCase versions 7.0.1.10 and
    7.1.1.3
    

Temporary fix

Comments

APAR Information

  • APAR number

    PM14503

  • Reported component name

    CLEARCASE UNIX

  • Reported component ID

    5724G2901

  • Reported release

    701

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    NoHIPER

  • Special Attention

    NoSpecatt

  • Submitted date

    2010-05-13

  • Closed date

    2010-08-19

  • Last modified date

    2010-08-19

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

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

Fix information

  • Fixed component name

    CLEARCASE UNIX

  • Fixed component ID

    5724G2901

Applicable component levels

  • R701 PSN

       UP

[{"Business Unit":{"code":"BU053","label":"Cloud & Data Platform"},"Product":{"code":"SSSH27","label":"Rational ClearCase"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"7.0.1","Edition":"","Line of Business":{"code":"LOB45","label":"Automation"}}]

Document Information

Modified date:
19 August 2010