And what does "ll_ost_io_**: page allocation failure" mean?
Nov 24 20:52:21 boss01 kernel: ll_ost_io_383: page allocation failure. order:4,
mode:0x50
Nov 24 20:52:21 boss01 kernel: [<c01425cc>] __alloc_pages+0x294/0x2a6
Nov 24 20:52:21 boss01 kernel: [<c01425f6>] __get_free_pages+0x18/0x24
Nov 24 20:52:21 boss01 kernel: [<c01451ed>] kmem_getpages+0x1c/0xbb
Nov 24 20:52:21 boss01 kernel: [<c0145d3b>] cache_grow+0xab/0x138
Nov 24 20:52:21 boss01 kernel: [<c0145f2d>]<4>ll_ost_io_137: page
allocation
failure. order:4, mode:0x50
Nov 24 20:52:21 boss01 kernel: [<c01425cc>]
cache_alloc_refill+0x165/0x19d
Nov 24 20:52:21 boss01 kernel: [<c0146301>] __alloc_pages+0x294/0x2a6
Nov 24 20:52:21 boss01 kernel: [<c01425f6>] __kmalloc+0x76/0x88
Nov 24 20:52:21 boss01 kernel: [<f9650369>] cfs_alloc+0x29/0x70 [libcfs]
Nov 24 20:52:21 boss01 kernel: [<fae1a75f>]
lprocfs_alloc_stats+0x11f/0x5b0
[obdclass]
Nikita Danilov ?:
> Wang lu writes:
> > Dear list,
>
> Hello,
>
> > There are some " failure to allocate a tage(XX)" in our
oss log, what
does> > this mean ?
>
> this means that the Lustre logging subsystem cannot allocate a page to
> store CDEBUG() messages. Often this means that messages are generated
> faster than debug_daemon manages to dump them to the log file. As a
> result, some parts of a debugging log will be lost, which is not
> especially important, unless you need the log to trace some problem.
>
> Nikita.