[dm-devel] RFC for multipath queue_if_no_path timeout.

[Mike Snitzer]

On Thu, Sep 26 2013 at  1:47pm -0400,
Frank Mayhar <fmayhar at google.com> wrote:

> On Thu, 2013-09-26 at 18:38 +0100, Alasdair G Kergon wrote:
> > On Thu, Sep 26, 2013 at 10:31:56AM -0700, Frank Mayhar wrote:
> > > Uh, huh.  And what about when (not if) _that_ fails?  (For one thing,
> > > what if the stuckness caused by the queued I/O prevents the binary from
> > > being successfully pulled in from storage?)
> >  
> > Lock the daemon in memory (or launch from ramdisk), don't allocate any new
> > memory while it's doing critical monitoring, tell the OOM killer not to kill
> > it, set high/real-time priority etc.
> > 
> > lvm2 and multipath-tools use some of these techniques and seem to cope OK.
> 
> Launching it from ramdisk won't help, particularly, since it still goes
> through the block layer.  The other stuff won't help if a (potentially
> unrelated) bug in the daemon happens to be being tickled at the same
> time, or if some dependency happens to be broken and _that's_ what's
> preventing the daemon from making progress.
> 
> And as far as lvm2 and multipath-tools, yeah, they cope okay in the kind
> of environments most people have, but that's not the kind of environment
> (or scale) we have to deal with.

Fine, please see the post I made earlier in this thread and let me know
what you think.