Re: Why add the general notification queue and its sources

From: David Lehman
Date: Thu Sep 05 2019 - 16:10:04 EST


On Thu, 2019-09-05 at 14:51 -0400, Ray Strode wrote:
> Hi,
>
> On Thu, Sep 5, 2019 at 2:37 PM Steven Whitehouse <swhiteho@xxxxxxxxxx
> > wrote:
> > The original reason for the mount notification mechanism was so
> > that we
> > are able to provide information to GUIs and similar filesystem and
> > storage management tools, matching the state of the filesystem with
> > the
> > state of the underlying devices. This is part of a larger project
> > entitled "Project Springfield" to try and provide better management
> > tools for storage and filesystems. I've copied David Lehman in,
> > since he
> > can provide a wider view on this topic.
> So one problem that I've heard discussed before is what happens in a
> thinp
> setup when the disk space is overallocated and gets used up. IIRC,
> the
> volumes just sort of eat themselves?
>
> Getting proper notification of looming catastrophic failure to the
> workstation user
> before it's too late would be useful, indeed.
>
> I don't know if this new mechanism dhowells has development can help
> with that,

My understanding is that there is already a dm devent that gets sent
when the low water mark is crossed for a thin pool, but there is
nothing in userspace that knows how to effectively get the user's
attention at that time.

> and/or if solving that problem is part of the Project Springfield
> initiative or not. Do you
> know off hand?

We have been looking into building a userspace event notification
service (for storage, initially) to aggregate and add context to low-
level events such as these, providing a single source for all kinds of
storage events with an excellent signal:noise ratio. Thin pool
exhaustion is high on the list of problems we would want to address.


David