On 08/07/2013 04:48 PM, emmanuel segura wrote:
This workaround is not for you, man mkfs.ocfs2 and look the option
-T filesystem-type
Specify how the filesystem is going to be used, so that
mkfs.ocfs2 can chose optimal filesystem parameters for that use. The
supported filesystem types
are:
mail Appropriate for file systems that will host
lots of small files.
datafiles
Appropriate for file systems that will host
a relatively small number of very large files.
vmstore
Appropriate for file systems that will host
Virtual machine images.
2013/8/7 "M.Atıf CEYLAN" <meh...@atifceylan.com
<mailto:meh...@atifceylan.com>>
Count: 243 Next Free Rec: 243
## Total Used Free Block#
0 4096 4096 0 7878657
1 4096 4096 0 11997697
2 4096 4095 1 12126721
3 4096 4095 1 12158977
4 4096 4094 2 933889
5 4096 <tel:5%C2%A0%C2%A0%C2%A0%204096> 4096
0 326657
6 4096 4096 0 327681
7 4096 4096 0 12417025
8 4096 4094 2 329729
9 4096 4091 5 1224193
10 4096 4090 6 1288705
11 4096 4089 7 1320961
12 4096 4096 0 1385473
13 4096 4096 0 1449985
Perhaps, I must increase the cluster size of ocfs2 :(
Is the discontig-bg a workaround for me? If it's a workaround I'll
make creation new volumes with big block size and move the data to
there.
What is your suggestion to me for mkfs.ocfs2 params? Journal size,
cluster size etc.?
(I have 6 volumes and each size is 140GB)
I have tried with the "-C 32K" parameter. But the result was bad than
before. http://www.novell.com/support/kb/doc.php?id=7008696 it says : "
In conjunction with fragmentation, the filesystem may run out of a
contiguous block to alloc for extent_alloc blocks leading to "out of
space" errors. This is an OCFS2 limitation, and the problem happens
*only when creating large files on file system with small Cluster
Size*." I have many small files (millions of files) on all volumes.
When I try with (http://oss.oracle.com/~seeda/misc/stat_sysdir.sh) this
script on old volumes I don't see any space problem.
some results of output of the script.
//global_bitmap
..................................
Bitmap Total: 44564302 Used: 27804345 Free: 16759957
Clusters per Group: 32256 Bits per Cluster: 1
Count: 243 Next Free Rec: 243
## Total Used Free Block#
0 193536 130058 63478 15676416
1 193536 120736 72800 23546880
2 193536 126088 67448 15740928
3 193536 130184 63352 15773184
4 193536 120517 73019 15805440
5 193536 115267 78269 31514112
6 193536 120907 72629 15869952
..................................................
Group Chain: 0 Parent Inode: 11 Generation: 3256184013
CRC32: 00000000 ECC: 0000
## Block# Total Used Free Contig Size
0 15676416 32256 15098 17158 503 4032
1 7838208 32256 31432 824 511 4032
2 3 32256 32256 0 0 4032
3 23514624 32256 17874 14382 446 4032
4 31352832 32256 15440 16816 507 4032
5 39191040 32256 17958 14298 479 4032
...................................................
//global_inode_alloc
Sub Alloc Slot: Global Sub Alloc Bit: 4
Bitmap Total: 269 Used: 32 Free: 237
Clusters per Group: 269 Bits per Cluster: 1
Count: 243 Next Free Rec: 1
## Total Used Free Block#
0 269 32 237 4
Group Chain: 0 Parent Inode: 8 Generation: 3256184013
CRC32: 00000000 ECC: 0000
## Block# Total Used Free Contig Size
0 4 269 32 237 237 4032
However I want to try with "--fs-features=discontig-bg" but I use
squeeze with 2.6.32 kernel. So this feature is not supported by the
running kernel (or I couldn't run it).
M.Atıf CEYLAN
Yurdum Yazılım
