Accordingly a lot of manuals - if you have block size 8KB and trying to write a 1KB file to the block, as result you waste 7KB of the block space. But recently I noticed about Fragments of File Block. In same case if you have File Block 8KB and Fragment size 1KB - you can save your block space, because 1KB file will fill only 1 fragment. Other 7 fragments will use for other files.
It seems like :
If you have large files - use fragment size equal to block size
If you have small files - use minimal fragment size
I want to know the best suggestion (relationship) between "File Block Size" and "Fragment Size".
I noticed that whole my root volume group has :
Block Size = 8K
Fragment Size = 8K
It's default setting. And it's strange because a lot of smallest file present in OS and I can waste more space than if I will use Fragment Size 1K.
Whenever you ask about block size, fragment size, and how blocks are allocated to files; you need to tell us what type of filesystem you're using. Different filesystems use different methods of allocating blocks for files.
I'm certainly not an expert on HP-UX filesystems, but I would suggest that you also specify the HP-UX release you're using to help volunteers here who may be able to help you respond appropriately.
Happy holidays...
That is the volume manager, but you probably mean VXFS? A fragment size equal to the block size makes sense for (cooked) database applications (especially if it is equal to the database block size) or for other applications where you mainly have large files. With lots of small files you are wasting space and a smaller fragment size would be more likely be the way to go.
Your root VG does not have a block size. Perhaps you mean that every VXFS filesystem in every LV in your root VG has a 8KB fragment size? That would not seem like a logical choice to me..
Oracle favors an 8KB fragment, indeed. In fact the Oracle installation guides for 11g say
exactly that.
It does not make sense to configure everything with 8kb.
The df -g command will list that information for you. Please share it with us.
$ df -g /
/ (/dev/vg00/lvol3 ) :
8192 file system block size 8192 fragment size
131072 total blocks 85350 total free blocks
85350 allocated free blocks 38944 total i-nodes
35436 total free i-nodes 35436 allocated free i-nodes
1073741827 file system id vxfs file system type
0x10 flags 255 file system name length
/ file system specific string
I think it's somehow related that root VG supposed to be on a contiguous LVM logical volume. But I'm not sure.
That does indeed show a "fragment size" of 8 KB, which is the "fundamental file system block size", i.e. the minimum block size that can be used on a file system..
It is important to get the terms right. A VG can not be on a LVM volume. What you mean is the root filesystem (in this case of type VXFS) that gets mounted on / . The root filesystem is usually on one of the logical volumes (typically lvol3) on vg00.
If there needs to be a contiguous LVM, then that typically pertains to the swap volume ( typically lvol2 on vg00), but that does not contain a VXFS file system and so does not have a block size / fragment size..). The contiguousness has to do with physical extents, not filesystem blocks / fragments..