Re: devfs persistence

From: Eduardo Horvath (eeh@turbolinux.com)
Date: Fri Apr 28 2000 - 15:04:07 EST


Hi Matt!

On Fri, 28 Apr 2000, Rodger Wilson {I/O Software} wrote:

> Subject: Re: devfs persistence
>
> What we do is refer to drives by world wide number, until it hits the
> Fiber channel HBA driver, at which time the WWN is exchanged for the
> current ALPA (target id) -- this way everything always works!
>
> Granted that refering to a drive as /dev/fc/w887384834,0 isn't pritty, but
> is reliabe, and then because this name is always the same you can create a
> symoblic link to it that is pritty and friendly.
>
> The catch is that the computer idustry in moving into Storage Area
> Networks (SANS -- probably knew that -- but I have been geting into
> trouble lattly). And all of these networkds (e.g. firewire, usb, fibre
> channel, infiniband...) all have some sort of dynaimic addressing
> mechanism. And if we are truely embrassing the idea of a SAN then we need
> to have friendly OS's on the SAN that don't write all over each others
> labels.

First of all Node or Port WWNs are not sufficient for this purpose. Let's
say you have a RAID box with two controllers. Each controllser has its
own WWN: WWN0 and WWN1. One of the controllers fails and needs to be
replaced. The new controller has a different WWN: WWN2. But it turns out
that the controller really wasn't bad, it just had a loose connection. So
it's used when a controller fails on another RAID box on the same
SAN. Now the original box has WWN1 and WWN2, but another box has WWN0 and
WWN3. The volumes are still in the original box, but now you have a new,
completely different set of volumes that magically appear attached to
WWN0.

No, the only reliable way to identify a platter is through the VIPD page.

> Imagion that in the future that there might be an ANSI standard filesystem
> and an ANSI standard partition, which are connected to a heterogenious
> SAN. But then NT goes and writes a label on the disk, then Solaris goes
> and over writes a small part of the NT label with it's own label, and then
> Linux comes along and over writes both of their labels with its own label.

An OS has no business overwriting a disk unless it is specifically
intructed to by the user. And if you do have multiple OSes on a SAN they
either understand each others' label formats and everything works or they
don't and couldn't share data that way in the first place.

The nice thing about using volume labels is that just about all major
volume/filesystem formats have space (sometimes unused) for a label. The
problem is getting the label onto an unlabeled device in the first place,
or trying to use a device without a label such as a tape.

Eduardo Horvath

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sun Apr 30 2000 - 21:00:15 EST