is there an autofs roadmap somewhere?

is there an autofs roadmap somewhere?

[Lever, Charles]

new to the game.

i'm wondering what features autofs developers have in mind for the near
future.

Re: is there an autofs roadmap somewhere?

[raven]

On Tue, 22 Jun 2004, Lever, Charles wrote:

> new to the game.
> 
> I'm wondering what features autofs developers have in mind for the near
> future.

Charles. Why did you have to ask "that" question!

There are quite a few things that people here on the list have asked for, 
that for one reason or another, are quite difficult to implement. I won't 
go there now.

The big issue for me is that, atm autofs does not have expected 
functionality in two areas. One is the lazy mounting of multi-mount maps 
and second, direct mounts that behave as expected (there's limited 
functionality now).

To be honest I've spent a lot of time fixing bugs. When that settles down 
(soon I hope) I want to re-work the code to sort out the above.

Once autofs has functionality expected of an automounter, a roadmap of 
additional features would be good to develop.

There's a redevelopment by Mike Waychison and Tim Hocking (of Sun) 
underway but they haven't delivered any product yet. Perhaps we'll see 
something in the 2.7 - 2.8 time frame. Their design will take much of the 
function into the kernel using callback helper utilities.

There's also the 800 mount limit in the sunrpc code for NFS v2 and v3 
mounts, but you know about that.

But what functionality do you need?
And what do you think would be useful?

Ian

RE: is there an autofs roadmap somewhere?

[Lever, Charles]

> On Tue, 22 Jun 2004, Lever, Charles wrote:
> 
> > new to the game.
> > 
> > I'm wondering what features autofs developers have in mind for the 
> > near future.
> 
> Charles. Why did you have to ask "that" question!

i'm working with trond to build a Linux NFS client roadmap.  our
customers want a little more transparency into the Linux NFS client
development plan, and i'm trying to include autofs because it is pretty
important for them.  i actually know very little about it (so a pointer
to a primer you like would be helpful for me).

> But what functionality do you need?
> And what do you think would be useful?

no requests... just polling for information, trying to get more involved
here.  thanks!

there are some naive thoughts about error recovery and threading
architecture that could help autofs deployments in very large
environments with lots of active mounts.  but i'd like to explore the
current autofs more before stepping in it.

RE: is there an autofs roadmap somewhere?

[raven]

On Fri, 25 Jun 2004, Lever, Charles wrote:

> > On Tue, 22 Jun 2004, Lever, Charles wrote:
> > 
> > > new to the game.
> > > 
> > > I'm wondering what features autofs developers have in mind for the 
> > > near future.
> > 
> > Charles. Why did you have to ask "that" question!
> 
> i'm working with trond to build a Linux NFS client roadmap.  our
> customers want a little more transparency into the Linux NFS client
> development plan, and i'm trying to include autofs because it is pretty
> important for them.  i actually know very little about it (so a pointer
> to a primer you like would be helpful for me).

Count me in.

Not sure what you need for doco here but the automounter howto at

http://www.linux-consulting.com/Amd_AutoFS/autofs.html

has a fair bit of info about autofs and amd, although a bit out of date.

But for a description of what's expected of an automounter the

"Managing NFS and NIS" by Stern, O'Reily, Ch 13

gives a good coverage, again a little incomplete these days.

> 
> > But what functionality do you need?
> > And what do you think would be useful?
> 
> no requests... just polling for information, trying to get more involved
> here.  thanks!
> 
> there are some naive thoughts about error recovery and threading
> architecture that could help autofs deployments in very large
> environments with lots of active mounts.  but i'd like to explore the
> current autofs more before stepping in it.

I've given some thought to a change in internal design. Introducing a 
threaded model in fact. But having lots of processes around and a 
cluttered mount table need to be considered, even with the current design.

I'd be interested in hearing your ideas about a threaded design.

Ian

Re: is there an autofs roadmap somewhere?

[Chris Walker]

Hear, hear.

We use autofs with lots of mounts, and we would appreciate any
development efforts in this area.

From Charles.Lever:
> > On Tue, 22 Jun 2004, Lever, Charles wrote:
> > 
> > > new to the game.
> > > 
> > > I'm wondering what features autofs developers have in mind for the 
> > > near future.
> > 
> > Charles. Why did you have to ask "that" question!
> 
> i'm working with trond to build a Linux NFS client roadmap.  our
> customers want a little more transparency into the Linux NFS client
> development plan, and i'm trying to include autofs because it is pretty
> important for them.  i actually know very little about it (so a pointer
> to a primer you like would be helpful for me).
> 
> > But what functionality do you need?
> > And what do you think would be useful?
> 
> no requests... just polling for information, trying to get more involved
> here.  thanks!
> 
> there are some naive thoughts about error recovery and threading
> architecture that could help autofs deployments in very large
> environments with lots of active mounts.  but i'd like to explore the
> current autofs more before stepping in it.
> 
> _______________________________________________
> autofs mailing list
> autofs@linux.kernel.org
> http://linux.kernel.org/mailman/listinfo/autofs

-- 
Chris Walker
Render Pipeline Group
Pixar Animation Studios
510/752.3736
cwalker@pixar.com

Re: is there an autofs roadmap somewhere?

[Ian Kent]

On Fri, 25 Jun 2004, Chris Walker wrote:

> Hear, hear.
> 
> We use autofs with lots of mounts, and we would appreciate any
> development efforts in this area.

And ...

RE: is there an autofs roadmap somewhere?

[Bryan O'Sullivan]

On Fri, 2004-06-25 at 23:18 +0800, raven@themaw.net wrote:

> I've given some thought to a change in internal design. Introducing a 
> threaded model in fact. But having lots of processes around and a 
> cluttered mount table need to be considered, even with the current design.

Tim Hockin and some other people at Sun are implementing an automounter
from scratch.  I don't recall much about the design, other than that it
seemed a bit inspired by hotplug (i.e. run a program to handle a mount/
unmount request).  The only reaction I've seen was from HPA, who wasn't
terribly impressed.

You can find the original thread here:

http://www.uwsg.iu.edu/hypermail/linux/kernel/0401.0/index.html#1415

	<b

-- 
Bryan O'Sullivan <bos@serpentine.com>

RE: is there an autofs roadmap somewhere?

[raven]

On Fri, 25 Jun 2004, Bryan O'Sullivan wrote:

> On Fri, 2004-06-25 at 23:18 +0800, raven@themaw.net wrote:
> 
> > I've given some thought to a change in internal design. Introducing a 
> > threaded model in fact. But having lots of processes around and a 
> > cluttered mount table need to be considered, even with the current design.
> 
> Tim Hockin and some other people at Sun are implementing an automounter
> from scratch.  I don't recall much about the design, other than that it
> seemed a bit inspired by hotplug (i.e. run a program to handle a mount/
> unmount request).  The only reaction I've seen was from HPA, who wasn't
> terribly impressed.
> 

Yes. Based on the proposal refered to there.

Additionally, they have developed a vfs expire extention to address the 
expiration requirements not covered in the original document.

Last time I asked Mike he said it was going fairly well.

They don't have clearance from Sun to release this atm so, as I said, not 
product as yet.

Ian