On Friday, 04/07/2006 at 10:30 AST, David Kreuter
<[EMAIL PROTECTED]> wrote:
> Seems to me CP already has most of this code path anyway. The free
> storage limit detection will stop a machine in its tracks after the
> second violation. It puts the errant machine into CP READ stopped state.
> In th
The shutdown/snapshot/restart method is what we use -- quick, solid, and
relatively painless. Still, I would prefer not to have to do it that way,
though. Doesn't linux maintain I/O performance heuristics that are lost at
shutdown time?
Jon
Linux has data in its
internal buffer cache that
> Well, personally, I think it would be better to address this totally
> from the Linux side. That is a "program" or script which could tell
all
> processes: Harden your data to disk and suspend processing. The
process
> would respond by hardening any data in its cache, then "do something"
so
> tha
Well, personally, I think it would be better to address this totally
from the Linux side. That is a "program" or script which could tell all
processes: Harden your data to disk and suspend processing. The process
would respond by hardening any data in its cache, then "do something" so
that it would
Pursuing this just a bit further down the rabbit hole...
I can see a couple of ways this could be usefully implemented. One way
would be the "no guest handshake expected" approach that just marks a
guest as "Do Not Dispatch" - more or less the equivalent of pushing the
STOP button on the hardwar
I agree. There are many ways of providing excellent HA solutions for
linux guest on z/VM, including redundant servers, sharing methods,
heartbeats, etc. Recall an earlier post about convincing the customers
do go with clustered solutions - as long as profit were there. The
solutions already exist.
Umm, not to be a pest, but aren't we kind of over-engineering here?
Coupled with LVM or EVMS snapshots, there are good in-guest backup
utilities easily available that make this problem trivially disappear,
and you get file-level backup and restore as a freebie. Do we really
need to teach CP the i
I *think* we're discussing two related, but separate, ideas here now:
1) freezing a guest so that it is no longer active (as PHSIII suggests,
this might not be all that difficult to do.)
2) "stunning" a guest (my term, at least) where the guest is put into a
state such that "interesting" things ca
Seems to me CP already has most of this code path anyway. The free
storage limit detection will stop a machine in its tracks after the
second violation. It puts the errant machine into CP READ stopped state.
In the distant past I made some presentations on this - I'll dig
around. How deep and tr
On 4/7/06, Phil Smith III <[EMAIL PROTECTED]> wrote:
> Well, that capability isn't that hard -- V/SPOOL (now part of
> CA-somethingorother-VM:Spool V/SEG-PLUS feature) has a FREEZE capability, and
> ISTR it was a few lines in the right place in CP to keep the guest from being
> dispatched.
>
I
Rob van der Heij <[EMAIL PROTECTED]> wrote:
>What we're missing is the ability to flash a set of mini disks as one
>atomic action. If you just go over the disks and flash them one at a
>time, in theory there is the risk of inconsistent backups. If you
>really want you can use LVM to suspend all wri
ue, N.W.
Washington, D.C. 20224
Voice: (202) 927-4188 FAX: (202) 622-6726
[EMAIL PROTECTED]
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of
Rob van der Heij
Sent: Thursday, April 06, 2006 5:38 PM
To: LINUX-390@VM.MARIST.EDU
Subject: Re: Freeze
> I am not sure the risk of inconsistent data in a backup is as real as
> some want you to believe.
Remind me to tell you about the conversation I had with the
chief sysprog many years ago when I said that it was pointless
backing up the warm & ckpt areas and he disagreed.
Shortly thereafter we h
On 4/6/06, Alan Altmark <[EMAIL PROTECTED]> wrote:
> If you could sync the disks and guarantee that it won't start writing
> again before you complete the flash, that would be ok, but I would not
> wish to depend on Divine Providence to protect the consistency of the
> backup.
What we're missing
On Wednesday, 04/05/2006 at 11:43 ZE2, Rob van der Heij
<[EMAIL PROTECTED]> wrote:
> On 4/5/06, Alan Altmark <[EMAIL PROTECTED]> wrote:
>
> > Image backups like that won't do what you want. Linux has data in its
> > internal buffer cache that needs to be written to dasd and that only
> > happens w
David Boyes wrote:
Dave Jones said:
I seem to recall that a number of years ago an IBM-er (Romney White,
perhaps) suggested a "stun" capability for VM. This would allow you to
stun a guest, do something to the system (maybe even move the guest to
another, similarly configured system) and have
Dave Jones said:
> I seem to recall that a number of years ago an IBM-er (Romney White,
> perhaps) suggested a "stun" capability for VM. This would allow you to
> stun a guest, do something to the system (maybe even move the guest to
> another, similarly configured system) and have it start back u
I believe you are thinking about the Single-System-Image program updates that
were done at University of Waterloo where Romney White worked. With enough CTCs
and shared DASD, multiple systems (like 4341s) could be used as a single system
or an unbalanced system where a user could be switched from s
On Wednesday, 04/05/2006 at 11:42 CST, Tyler Koyl <[EMAIL PROTECTED]>
wrote:
>> I am quite sure this exists but I need a little help.
>> I was wondering if there is a way to place a z/VM 5.1 linux guest into a
froozen
>> state? ...
And on Wednesday 05 April 2006 15:48, Alan Altmark responded:
>I
On 4/5/06, Alan Altmark <[EMAIL PROTECTED]> wrote:
> Image backups like that won't do what you want. Linux has data in its
> internal buffer cache that needs to be written to dasd and that only
> happens when you unmount the filesystem from Linux or shut Linux down.
There's one of the magic SysR
I recall that at one time somebody set up something like that to move CMS users
from one processor to another. I don't remember much details but it seem like it
required a shared directory and maybe a shared spool between the systems. This
was back in the late 80's.
[EMAIL PROTECTED] wrote:
I b
SD and are linked via Gigabit ethernet. It's pretty slick.
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of
Dave Jones
Sent: Wednesday, April 05, 2006 4:15 PM
To: LINUX-390@VM.MARIST.EDU
Subject: Re: Freeze Linux Guests momentarily
VM Ware o
half Of
Tyler Koyl
Sent: Wednesday, April 05, 2006 10:42 AM
To: LINUX-390@VM.MARIST.EDU
Subject: [LINUX-390] Freeze Linux Guests momentarily
I am quite sure this exists but I need a little help.
I was wondering if there is a way to place a z/VM 5.1 linux guest into a
froozen state? My thinking i
.EDU
390 Portcc
<[EMAIL PROTECTED]
T.EDU> Subject
Re: Freeze Linux Guests momentarily
04/05/2006 12:38 P
On Wednesday, 04/05/2006 at 11:42 CST, Tyler Koyl <[EMAIL PROTECTED]>
wrote:
> I am quite sure this exists but I need a little help.
>
> I was wondering if there is a way to place a z/VM 5.1 linux guest into a
froozen
> state? My thinking is just place the guest into a CP Read State. Can't
seem to
y, April 05, 2006 10:42 AM
To: LINUX-390@VM.MARIST.EDU
Subject: [LINUX-390] Freeze Linux Guests momentarily
I am quite sure this exists but I need a little help.
I was wondering if there is a way to place a z/VM 5.1 linux guest into a
froozen state? My thinking is just place the guest into a CP Re
tely by reply e-mail
and delete this message. Thank you for your cooperation."
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of
Dave Jones
Sent: Wednesday, April 05, 2006 11:28 AM
To: LINUX-390@VM.MARIST.EDU
Subject: Re: [LINUX-390] Freeze Linux Guests m
AM
To: LINUX-390@VM.MARIST.EDU
Subject: [LINUX-390] Freeze Linux Guests momentarily
I am quite sure this exists but I need a little help.
I was wondering if there is a way to place a z/VM 5.1 linux guest into a
froozen state? My thinking is just place the guest into a CP Read State.
Can't seem to
uot;
-Original Message-
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of
Tyler Koyl
Sent: Wednesday, April 05, 2006 10:42 AM
To: LINUX-390@VM.MARIST.EDU
Subject: [LINUX-390] Freeze Linux Guests momentarily
I am quite sure this exists but I need a little help.
I was wondering
I am quite sure this exists but I need a little help.
I was wondering if there is a way to place a z/VM 5.1 linux guest into a froozen
state? My thinking is just place the guest into a CP Read State. Can't seem to
find the command to send a guest a request to enter CP READ.
I am writing a REXX ex
30 matches
Mail list logo
