Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 18 Jul 2005 01:14:39 +0200 (CEST)
From:      Ivan Voras <ivoras@fer.hr>
To:        Giorgos Keramidas <keramida@freebsd.org>
Cc:        freebsd-current@freebsd.org
Subject:   Re: Errno man page
Message-ID:  <20050718004649.U70085@geri.cc.fer.hr>
In-Reply-To: <20050717223729.GD1291@gothmog.gr>
References:  <20050718000738.F69475@geri.cc.fer.hr> <20050717223729.GD1291@gothmog.gr>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 18 Jul 2005, Giorgos Keramidas wrote:

> On 2005-07-18 00:14, Ivan Voras <ivoras@fer.hr> wrote:
>> I think the errno(i.e. intro(2)) page needs to be updated. There is
>> some information that doesn't "feel" current:
>>
>> - for EFBIG (#27) - I hope the limit is > 2.1E9 on ufs2 :)
>

> I think that instead of trying to guess a value that would be correct for
> many filesystems, but obviously wrong for others, we should just remove the
> explicit size.

Agreed, but since UFS2 is The Filesystem for FreeBSD for the forseeable 
future, it makes sense to state something like "the limit is X on UFS2, 
and can vary from filesystem to filesystem".

>
>> - for EMFILE - is the limit on open files really 64 per process?
> system that is a little lower than kern.maxfiles:

Yes, that's what I was aiming at :) Maybe the sysctl deserves to be 
mentioned in the man page?

>> (And of course, tuning(7) also has some historical figures)
>
> Can you help us identify them?

I remember this discussed some time ago so I assumed it's fixed, but 
here are some suggestions:

- the default partition sizes at the start of the page are a bit low (but 
nothing serious) - maybe just double the sizes for root & /var.
- About the swap size: """The kernel's VM paging algorithms are tuned to 
perform best when there is at least 2x swap versus main 
memory.  Configuring too little swap can lead to inefficiencies in the 
VM""" - AFAIK this is not really true? - also, with 64bit computers 
becoming as common as 32bit, maybe a note should be added about the 
differences (also AFAIK: there's no point in having RAM+swap > 4GB on 
32bit machines?)
- maybe a note should be added about the need for large amounts of memory 
for running fsck on TB-sized filesystems?
- I don't know if information in kern.maxusers and kern.ipc.nmbclusters 
sysctls is correct, but I seem to recall some discusstions where they 
ended up higher than the recommended maximums in the man pages.

Actually, there aren't many errors, sorry for the false alarm :)

-- 
Every sufficiently advanced magic is indistinguishable from technology
    - Arthur C Anticlarke




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050718004649.U70085>