Re: "Swap area shorter than signature indicates"
On Mon, May 07, 2007 at 09:24:18PM +0200, Matteo Vescovi wrote:
> The subject of this mail is the message I can find for every boot in my
> /var/log/messages file.
> I've set up a RAID1 environment with etch amd64; it's composed of 2
> identical HDDs. The partitioning scheme is the same for both the devices.
> I've put the swap area (a /dev/md1 made by /dev/sda5 and /dev/sdb5) in
> the RAID, setting the ID label as "fd" (that is the 'Linux raid
> autodetect' mode).
>
> The /proc/mdstat file says about that partition:
>
> md1 : active raid1 sda5[0] sdb5[1]
> 1951744 blocks [2/2] [UU]
>
> Is that correct?
That message is what to expect for a raid1 with both devices up and
running (U = Up)
> Is the swap area working right despite the message above?
> If so, why do I get that strange message in the logs?
Did you mkswap /dev/md1 or /dev/sda5?
Try doing swapoff, then make sure free shows no swap space, then use
mkswap to recreate the swap on /dev/md1 and then swapon again and see if
it gives a message then. Making the swap on the low level filesystem
rather than the raid one would cause such problems, as would taking an
existing swap partition and adding a new drive and making raid on it and
not reinitializing the swap space to match the new slightly smaller size
(since raid does require a small superblock to store it's information).
--
Len Sorensen
Reply to: