Subject: Re: sysinst error: uid 0 comm newfs on /:file system full
To: Gerardo Alvarez =?iso-8859-1?Q?Le=F3n?= <gerardo.alvarez@wanadoo.es>
From: Frederick Bruckman <fredb@immanent.net>
List: port-mac68k
Date: 08/16/2002 19:01:04
On Fri, 16 Aug 2002, Gerardo Alvarez Le=F3n wrote:

> Warning: 727 sector(s) in last cylinder unallocated
> /dev/rwd0a:=096053888 sectors in 6407 cylinders of 15 tracks, 63 sectors
> uid 0 comm newfs on /:file system full
>
> /: write failed, file system is full
>
> Command
> =09/sbin/newfs /dev/rwd0a failed. I can=B4t continue.
>
> I exit and do a df and this is what it shows:
> Filesystem=09512-blocks=09Used=09Avail=09Capacity=09Mounted on
> /dev/md0a=09      3967=093106=09  861=09     78%=09/
> kernfs=09                 2=09   2=09    0=09    100%=09/kern
>
> I am downloading snapshot 1.6F to try it but I am afraid it will give
> the same error...
> Any hint or workaround?
> Which file system is full? Or is it a "fake" error?

The error's actually from the kernel, not from "newfs". It's telling
you why it killed your task in mid-stream.

The ramdisk may be out of inodes, even with blocks available, or
it may have actually run out of space while executing "newfs", even
though it's back when you look, after "newfs" has died".

The best solution is probably to bump the size of the ramdisk.

Here's a crazy idea: Assuming the partitions have already been made,
escape to the shell and try adding swap before making the file system
manually. Something like

1) Type "x" to exit sysinstall,

2)
=09disklabel wd0
=09# (review the partitions for correctness)
=09swapctl -a /dev/wd0b
=09newfs /dev/rwd0a

3) Type Control-D to restart "sysinstall".


Frederick