[Linux-cluster] lock_gulm is very slow. why ?

[Levent Serinol]

Hi,
 I have done some benchmark tests with postmark(tests repeated many
times). There is one client (also it is lock server). and another one
which exports it's scsi hard disk with gnbd.

filesystem created with lock_gulm:
-----------------------------------------
Time:
        94 seconds total
        7 seconds of transactions (142 per second)

Files:
        10692 created (113 per second)
                Creation alone: 10000 files (434 per second)
                Mixed with transactions: 692 files (98 per second)
        899 read (128 per second)
        101 appended (14 per second)
        10692 deleted (113 per second)
                Deletion alone: 10384 files (162 per second)
                Mixed with transactions: 308 files (44 per second)

Data:
        21.05 megabytes read (229.28 kilobytes per second)
        250.41 megabytes written (2.66 megabytes per second)


filesystem created with no_lock:
--------------------------------------
Time:
        35 seconds total
        4 seconds of transactions (250 per second)

Files:
        10692 created (305 per second)
                Creation alone: 10000 files (454 per second)
                Mixed with transactions: 692 files (173 per second)
        899 read (224 per second)
        101 appended (25 per second)
        10692 deleted (305 per second)
                Deletion alone: 10384 files (1153 per second)
                Mixed with transactions: 308 files (77 per second)

Data:
        21.05 megabytes read (615.77 kilobytes per second)
        250.41 megabytes written (7.15 megabytes per second)


as you can see nolock results is 2 times (some parts 3 times) faster
then with locked one .
what could be the problem ? is there any workaround or settune option
(releasing locks earlier,etc...) ?