Re: newbie question - convert to full pack minidisk

2008-04-27 Thread Quay, Jonathan (IHG)
I would consider the system thrashing due to dramatic overcommitment of storage 
by minidisk cache when taking IBM provided mdc arbiter defaults to be a bug.



From: The IBM z/VM Operating System on behalf of Rob van der Heij
Sent: Sat 4/26/2008 5:19 AM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: newbie question - convert to full pack minidisk



On Fri, Apr 25, 2008 at 10:25 PM, Quay, Jonathan (IHG)
<[EMAIL PROTECTED]> wrote:

> We found a bug in CP minidisk cacheing that would lock up our system.  We 
> turn off MDC at the device level in the system configuration file for our 
> guest operating system minidisks using RDEV  TYPE DASD MDC OFF.

It would take real performance data to conclude on a bug.
It may just be tuning. We know that the defaults for the MDC arbiter
are such that many systems will favor MDC and thus cause more paging
than the MDC hit rate justifies. If the paging subsystem is not
properly configured, then that may keep folks waiting or even abend
the system. It's a very popular problem at new installations.
When the I/O is not eligable for MDC, the arbiter gets no opportunity
to favor that, and you avoid the paging.
I would prefer to properly configure MDC rather than disable it. And
make sure the paging subsystem is able to handle the load when
something else makes you page.

Rob
--
Rob van der Heij
Velocity Software GmbH
http://velocitysoftware.com/


Re: newbie question - convert to full pack minidisk

2008-04-26 Thread Rob van der Heij
On Fri, Apr 25, 2008 at 10:25 PM, Quay, Jonathan (IHG)
<[EMAIL PROTECTED]> wrote:

> We found a bug in CP minidisk cacheing that would lock up our system.  We 
> turn off MDC at the device level in the system configuration file for our 
> guest operating system minidisks using RDEV  TYPE DASD MDC OFF.

It would take real performance data to conclude on a bug.
It may just be tuning. We know that the defaults for the MDC arbiter
are such that many systems will favor MDC and thus cause more paging
than the MDC hit rate justifies. If the paging subsystem is not
properly configured, then that may keep folks waiting or even abend
the system. It's a very popular problem at new installations.
When the I/O is not eligable for MDC, the arbiter gets no opportunity
to favor that, and you avoid the paging.
I would prefer to properly configure MDC rather than disable it. And
make sure the paging subsystem is able to handle the load when
something else makes you page.

Rob
-- 
Rob van der Heij
Velocity Software GmbH
http://velocitysoftware.com/


Re: newbie question - convert to full pack minidisk

2008-04-25 Thread dave
Really? That's interestingwhat version of z/VM are you
running there? And I take it that the guest operating
systems are Linux, right?

DJ

- Original Message Follows -
From: "Quay, Jonathan (IHG)" <[EMAIL PROTECTED]>
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: newbie question - convert to full pack minidisk
Date: Fri, 25 Apr 2008 16:25:22 -0400

> We found a bug in CP minidisk cacheing that would lock up
> our system.  We turn off MDC at the device level in the
> system configuration file for our guest operating system
> minidisks using RDEV  TYPE DASD MDC OFF.
> 
> 
> 
> From: The IBM z/VM Operating System on behalf of Paul
> Vincent Sent: Fri 4/25/2008 3:04 PM
> To: IBMVM@LISTSERV.UARK.EDU
> Subject: newbie question - convert to full pack minidisk
> 
> 
> 
> All,
> 
> I would like to convert a Mod-9 defined as follows:  MDISK
> 0300 3390 0001 end XX
> to a full pack minidisk as follows: MDISK 0300 3390 
> end XX. without losing any data  (the volume contains
> Linux root filesystem). Can this be done?  if so how (DDR,
> FLAHSCOPY?) If I understand what I've read, I should get
> some performance from a full pack minidisk (not as much CP
> overhead) and I won't have to code a MINIOPT NOMDC
> statement I'm I on the right track?
> 
> Thanks in advance.
> 
> Paul
> 
> 
> 
> --
> --- This message, together with any attachments,
> is intended only for the use of the individual or entity
> to which it is addressed. It may contain information
> that is confidential and prohibited from disclosure.
> If you are not the intended recipient, you are hereby
> notified that any dissemination or copying of this
> message or any attachment is strictly prohibited. If
> you have received this message in error, please notify
> the original sender immediately by telephone or by
> return e-mail and delete this message, along with any
> attachments, from your computer. Thank you.
> 
> --
> ---


Re: newbie question - convert to full pack minidisk

2008-04-25 Thread Quay, Jonathan (IHG)
We found a bug in CP minidisk cacheing that would lock up our system.  We turn 
off MDC at the device level in the system configuration file for our guest 
operating system minidisks using RDEV  TYPE DASD MDC OFF.



From: The IBM z/VM Operating System on behalf of Paul Vincent
Sent: Fri 4/25/2008 3:04 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: newbie question - convert to full pack minidisk



All,

I would like to convert a Mod-9 defined as follows:  MDISK 0300 3390 0001
end XX
to a full pack minidisk as follows: MDISK 0300 3390  end XX.
without losing any data  (the volume contains Linux root filesystem).
Can this be done?  if so how (DDR, FLAHSCOPY?)
If I understand what I've read, I should get some performance from a full
pack minidisk (not as much CP overhead) and I won't have to code a MINIOPT
NOMDC statement
I'm I on the right track?

Thanks in advance.

Paul



-
This message, together with any attachments, is
intended only for the use of the individual or entity
to which it is addressed. It may contain information
that is confidential and prohibited from disclosure.
If you are not the intended recipient, you are hereby
notified that any dissemination or copying of this
message or any attachment is strictly prohibited. If
you have received this message in error, please notify
the original sender immediately by telephone or by
return e-mail and delete this message, along with any
attachments, from your computer. Thank you.

-


Re: newbie question - convert to full pack minidisk

2008-04-25 Thread Rob van der Heij
On Fri, Apr 25, 2008 at 9:04 PM, Paul Vincent
<[EMAIL PROTECTED]> wrote:
> All,
>
>  I would like to convert a Mod-9 defined as follows:  MDISK 0300 3390 0001
>  end XX
>  to a full pack minidisk as follows: MDISK 0300 3390  end XX.
>  without losing any data  (the volume contains Linux root filesystem).
>  Can this be done?  if so how (DDR, FLAHSCOPY?)

Since you copy in the right direction, I think you should be able to
DDR with REORDER 0
to the same volume. But I would never dare and would DDR to a new volume.

>  If I understand what I've read, I should get some performance from a full
>  pack minidisk (not as much CP overhead) and I won't have to code a MINIOPT
>  NOMDC statement

But then you have to code real device addresses in the CP directory
with all kind of options to get it wrong.
If you want to avoid MDC you might also tell CP in the system
configuration file that the devices are shared... but in general I
think MDC is goodness.

I do not think you will be able to measure the difference unless there
is a bug in CP...

>  I'm I on the right track?

Wrong cylinder, I would say ;-)

-- 
Rob van der Heij
Velocity Software GmbH
http://velocitysoftware.com/


newbie question - convert to full pack minidisk

2008-04-25 Thread Paul Vincent
All,

I would like to convert a Mod-9 defined as follows:  MDISK 0300 3390 0001
end XX
to a full pack minidisk as follows: MDISK 0300 3390  end XX.
without losing any data  (the volume contains Linux root filesystem).
Can this be done?  if so how (DDR, FLAHSCOPY?)
If I understand what I've read, I should get some performance from a full
pack minidisk (not as much CP overhead) and I won't have to code a MINIOPT
NOMDC statement
I'm I on the right track?

Thanks in advance.

Paul



-
This message, together with any attachments, is
intended only for the use of the individual or entity
to which it is addressed. It may contain information
that is confidential and prohibited from disclosure.
If you are not the intended recipient, you are hereby
notified that any dissemination or copying of this
message or any attachment is strictly prohibited. If
you have received this message in error, please notify
the original sender immediately by telephone or by
return e-mail and delete this message, along with any
attachments, from your computer. Thank you.

-