[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Mail Server Problem



Yes 

The block size is 8k below is the output from fstyp 

bsize   8192

So what I hear you say is we re-initialize the disk with 4k bsize and 2k
inode size? And that way we get the best use out of the disk for the
maildir storage area.

If I understand this well (nifree) is the number of free inodes and with
the previous disk setup we were kind of okay, but (nbfree) which is the
number of free blocks was very small and so we could not create large
files on the disk.

Thus the solution would have been to reinitiallize the disk with the
default inode size ( -i 2048) and a smaller block size (-b 4096).

I will try it again with the reduced block size.

--Ayitey 


-----Original Message-----
From: owner-afnog at afnog.org [mailto:owner-afnog at afnog.org] On Behalf Of
Brian Candler
Sent: Friday, January 03, 2003 5:42 PM
To: Ayitey Bulley
Cc: 'M. K. Agbessitse'; 'Mensah Komla Agbessitse'; afnog at afnog.org;
brunomake at yahoo.co.uk
Subject: Re: Mail Server Problem


On Fri, Jan 03, 2003 at 05:22:02PM -0000, Ayitey Bulley wrote:
> Bruno,
> 
> Actually the default for Solaris I believe is 2k so 1k just doubles 
> the number of inodes.

Ah, then we are talking about two different things.

There were plenty of inodes before, so creating more (e.g. one inode per
1k instead of one inode per 2k) is not the solution.

What I was after was the block size (displayed as 'bsize' by fstyp),
which is 8K. I have access to a Solaris machine at the moment; man newfs
says

               -b bsize  The logical block size of the file  sys-
                         tem  in bytes (either 4096 or 8192). The
                         default is 8192. The sun4u  architecture
                         does not support the 4096 block size.

That's what I was talking about. Try it again with '-b 4096'

The strange thing about the Sun filesystem is that there can be quite a
lot of space available as shown by 'df', but it's actually unable to
create a file, because if it wants to write 8K of data it has to sit in
a whole block; it cannot be spread across free fragments in multiple
blocks.

I think that what Bruno has done has rebuilt the filesystem with the
same bsize, and has gained some space as a result of it being less
fragmented: 172K blocks, or 172*8 = 1.3GB of space.

However, over time it will need defragmenting again and again, which is
not a good way to run a mailserver. So it's probably still worth a newfs
with -b 4096.

The option for inodes is:

               -i nbpi   The number  of  bytes  per  inode.  This
                         specifies  the  density of inodes in the
                         file system. The number is divided  into
                         the  total  size  of  the file system to
                         determine the fixed number of inodes  to
                         create.  It  should reflect the expected
                         average size of files in the  file  sys-
                         tem.  If  fewer  inodes  are  desired, a
                         larger number should be used; to  create
                         more  inodes  a smaller number should be
                         given. The default is 2048.

Regards,

Brian.

> 
> 
> --Ayitey
> 
> My mail server problem is now solved. We have used the option of 1k 
> (1024bytes), which according to the last mail received from you 
> implies that we have 8 times original format. # df -k
> Filesystem            kbytes    used   avail capacity  Mounted on
> /dev/dsk/c0t11d0s2   15683444 14030051 1496559    91%    /home
> #
> 
> And I still have more nbfree now.
> # fstyp -v /dev/dsk/c0t11d0s2 | head -20 | grep nbfree
> nbfree  172405  ndir    43819   nifree  14689593        nffree  272754
> Broken Pipe
> Unknown_fstyp (no matches)
> #
> 
> Thank You very much.
> 
> Bruno.
> 
> 
> -----
> This is the afnog mailing list, managed by Majordomo 1.94.5
> 
> To send a message to this list, e-mail afnog at afnog.org
> To send a request to majordomo, e-mail majordomo at afnog.org and put 
> your request in the body of the message (i.e use "help" for help)
> 
> This list is maintained by owner-afnog at afnog.org
> 
> 
> 
> -----
> This is the afnog mailing list, managed by Majordomo 1.94.5
> 
> To send a message to this list, e-mail afnog at afnog.org
> To send a request to majordomo, e-mail majordomo at afnog.org and put 
> your request in the body of the message (i.e use "help" for help)
> 
> This list is maintained by owner-afnog at afnog.org
> 
> 

-----
This is the afnog mailing list, managed by Majordomo 1.94.5

To send a message to this list, e-mail afnog at afnog.org
To send a request to majordomo, e-mail majordomo at afnog.org and put your
request in the body of the message (i.e use "help" for help)

This list is maintained by owner-afnog at afnog.org



-----
This is the afnog mailing list, managed by Majordomo 1.94.5

To send a message to this list, e-mail afnog at afnog.org
To send a request to majordomo, e-mail majordomo at afnog.org and put
your request in the body of the message (i.e use "help" for help)

This list is maintained by owner-afnog at afnog.org