* Xen watchdog?
@ 2007-09-25 8:37 Richard W.M. Jones
2007-09-25 9:27 ` Keir Fraser
0 siblings, 1 reply; 3+ messages in thread
From: Richard W.M. Jones @ 2007-09-25 8:37 UTC (permalink / raw)
To: xen-devel
[-- Attachment #1.1: Type: text/plain, Size: 586 bytes --]
I noticed that a Xen watchdog implementation was proposed on this list a
few months back:
http://lists.xensource.com/archives/html/xen-devel/2007-05/msg01159.html
Has this been replaced with something else? If not then did anything
come of this and/or are there any plans for a watchdog in a future
hypervisor?
Rich.
--
Emerging Technologies, Red Hat - http://et.redhat.com/~rjones/
Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod
Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in
England and Wales under Company Registration No. 03798903
[-- Attachment #1.2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3237 bytes --]
[-- Attachment #2: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Xen watchdog?
2007-09-25 8:37 Xen watchdog? Richard W.M. Jones
@ 2007-09-25 9:27 ` Keir Fraser
2007-09-25 17:34 ` Mark Williamson
0 siblings, 1 reply; 3+ messages in thread
From: Keir Fraser @ 2007-09-25 9:27 UTC (permalink / raw)
To: Richard W.M. Jones, xen-devel; +Cc: Mark Williamson
I took a brief look at it. It was kind of big, with no great justification.
I would have written a simple little user daemon for kicking the watchdog
directly, rather than add extra stuff in our Linux kernel. But perhaps the
latter is the 'proper' way? If so, the driver probably needs review on lkml.
Either way, I think the hypervisor interface should support multiple
entities per guest trying to set up a watchdog timer. Even if just so that
entities that lose the race get -EBUSY. This would make it safer to use
without a single controlling entity in the guest kernel.
Are you up for kicking these patches into shape, Mark?
-- Keir
On 25/9/07 09:37, "Richard W.M. Jones" <rjones@redhat.com> wrote:
> I noticed that a Xen watchdog implementation was proposed on this list a
> few months back:
>
> http://lists.xensource.com/archives/html/xen-devel/2007-05/msg01159.html
>
> Has this been replaced with something else? If not then did anything
> come of this and/or are there any plans for a watchdog in a future
> hypervisor?
>
> Rich.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Xen watchdog?
2007-09-25 9:27 ` Keir Fraser
@ 2007-09-25 17:34 ` Mark Williamson
0 siblings, 0 replies; 3+ messages in thread
From: Mark Williamson @ 2007-09-25 17:34 UTC (permalink / raw)
To: Keir Fraser; +Cc: xen-devel, Richard W.M. Jones
> I took a brief look at it. It was kind of big, with no great justification.
> I would have written a simple little user daemon for kicking the watchdog
> directly, rather than add extra stuff in our Linux kernel. But perhaps the
> latter is the 'proper' way? If so, the driver probably needs review on
> lkml.
It's the Linux Way - a userspace app wouldn't work, since it wouldn't be able
to export the /dev/watchdog device node that HA infrastructure needs.
To be honest, IIRC, it seemed to me like most watchdog drivers in Linux could
share more common code and then all be a decent bit smaller - something I
could look at if the Linux folks are interested but wasn't on my todo list at
the time.
A fair bit of the bulk in the rest of the code was in making the dom0 tools be
a bit more aware of the watchdog state. This was needed to support features
such as the "watchdog initiated reboot" flag, but isn't strictly needed for
basic functionality. I can trim this bit out, at least for an initial merge,
but there may be other stuff it'd be helpful to hook in here.
> Either way, I think the hypervisor interface should support multiple
> entities per guest trying to set up a watchdog timer. Even if just so that
> entities that lose the race get -EBUSY. This would make it safer to use
> without a single controlling entity in the guest kernel.
Hmmm. I could do something like that, certainly. The model I was following
was that of hardware - I'm not sure what a physical watchdog card would do
with multiple controlling entities but I doubt it'd be pretty. As we're in
the hypervisor, we've got more smarts available to us though, so a bit of
opaque state to identify a "watchdog handle" should be trivial to add. I
wonder if there are any potential users this sort of behaviour, though -
maybe Richard would comment here?
The only significant feature I'd want to add before merge is support for
suspend / resume - that should be fairly simple to add, though.
Cheers,
Mark
--
Dave: Just a question. What use is a unicyle with no seat? And no pedals!
Mark: To answer a question with a question: What use is a skateboard?
Dave: Skateboards have wheels.
Mark: My wheel has a wheel!
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2007-09-25 17:34 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-09-25 8:37 Xen watchdog? Richard W.M. Jones
2007-09-25 9:27 ` Keir Fraser
2007-09-25 17:34 ` Mark Williamson
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.