Re: PP size: 32MB vs. 64MB

From: Dan Foster (dsf_at_globalcrossing.net)
Date: 10/09/03 

Date: Thu, 9 Oct 2003 14:42:42 +0000 (UTC)

In article <b8cdc181.0310090630.8253733@posting.google.com>, Chris <krzgru@yahoo.co.uk> wrote:
>
> Please tell me what is the relation between inode "usage" on
> filesystem made on LV created one on VG with PP 32MB size and second
> on VG with PP 64MB PP size.

I don't think there's any direct relationship between the two.

PP is just a chunk of disk used to construct the underlying logical volume;
that's a lower layer than the filesystem which is constructed from a
logical volume.
 
> AIX 4.3.3, PP 32MB
> Filesystem 512-blocks Free %Used Iused %Iused Mounted on
> /dev/datalv 10354688 4285336 59% 666869 52% /datalv
>
> AIX 5.2, PP 64MB
> Filesystem 512-blocks Free %Used Iused %Iused Mounted on
> /dev/datalv 10485760 2354800 78% 927235 71% /datalv
>
> The files which were transfered are small files.
>
> Why having the same data in both FS, used space and inode numbers are
> different?

Could be a jfs vs jfs2 difference. You may want to post the output of:

$ lsfs -q /datalv

from both systems.

It could also be crfs/mkfs with different logic in selecting default
settings; different in 4.3.3 from 5.2?

> Is it PP size related? How can i increase inode number - only by
> increasing the FS size?

To increase inode count, you'd have to back up the filesystem, wipe it,
redo the mkfs (or crfs) command to recreate the filesystem with a different
-o nbpi=xxxx option (and you might need to adjust ag and frag in some cases
if required to match the nbpi setting). Then once filesystem created to
your satisfaction, restore the data from backup. (From tape, a tar file, a
mksysb, or whatever you like.)

Also, 'man crfs' for more info on nbpi and other settings.

-Dan

Relevant Pages

  • Re: 6.x loosing record of free space after filesystem fills?
    ... These settings were chosen to optimize I/O throughput for Postgresql on the theory that a 64KB block size would maximize disk throughput in the general case and an 8K frag size would match Postgresql's page size. ... I wasn't aware of any known regressions in 6.x regarding large filesystem block sizes... ... > within the filesystem code and VM code that will cause such filesystems ...
    (freebsd-stable)
  • Re: NFS and partitioned md
    ... Everything worked with same settings, ... switching my multiple md devices into md_d. ... All it cares about is that the type of filesystem is supported. ...
    (Linux-Kernel)
  • Re: [opensuse] Converting file system
    ... logbufs is bsize/sectsz defaultYou find the ... settings with 'xfs_info filesystem'. ...
    (SuSE)
  • Re: SetTimeZoneInformation and file time stamps
    ... It's entirely up to the filesystem for a given piece of hardware how times ... They could be stored in local time, rather than GMT, for ... possible to assure that the time zone doesn't change between copying a file ... settings are the same, you should get a good compare. ...
    (microsoft.public.windowsce.platbuilder)
  • Re: mkfs or crfs
    ... Subject: mkfs or crfs ... JFSlog in general which you might find interesting. ... there's a lot of I/O activity to a filesystem that doesn't mean that the ...
    (AIX-L)