I have marked a number of CPUs in a machine no-intr using psradm.
I can't see that this has changed anything on disk anywhere, so I am
slightly worried that this change will not persist across a reboot.
Will it? I am running Solaris 10, and the manuals are drawing a blank.
If not, is there a canonical place to put this initialisation?
Ceri
--
That must be wonderful! I don't understand it at all.
-- Moliere
|
|
0
|
|
|
|
Reply
|
Ceri
|
4/11/2006 1:58:15 PM |
|
On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
> I have marked a number of CPUs in a machine no-intr using psradm.
> I can't see that this has changed anything on disk anywhere, so I am
> slightly worried that this change will not persist across a reboot.
>
> Will it? I am running Solaris 10, and the manuals are drawing a blank.
The answer to that question is "no".
> If not, is there a canonical place to put this initialisation?
Still open!
Ceri
--
That must be wonderful! I don't understand it at all.
-- Moliere
|
|
0
|
|
|
|
Reply
|
Ceri
|
4/11/2006 5:12:23 PM
|
|
Ceri Davies <ceri_usenet@submonkey.net> writes:
> On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
>> I have marked a number of CPUs in a machine no-intr using psradm.
>> I can't see that this has changed anything on disk anywhere, so I am
>> slightly worried that this change will not persist across a reboot.
>>
>> Will it? I am running Solaris 10, and the manuals are drawing a blank.
>
> The answer to that question is "no".
>
>> If not, is there a canonical place to put this initialisation?
>
> Still open!
Hi Ceri :-)))
I'm not sure about a "canonical" place, but an /etc/init.d script that
runs before important stuff fire up (i.e. network interfaces) seems ok.
If you feel funky, an SMF service somewhere after milestone/single-user
would be nice too.
|
|
0
|
|
|
|
Reply
|
Giorgos
|
4/11/2006 10:53:58 PM
|
|
Perhaps psrset and psradm should be persistent?
What are your thoughts on that?
- Bart
|
|
0
|
|
|
|
Reply
|
bart
|
4/12/2006 4:39:09 AM
|
|
In article <1144816749.555175.3780@u72g2000cwu.googlegroups.com>, bart.smaalders@gmail.com <bart.smaalders@gmail.com> wrote:
>
> Perhaps psrset and psradm should be persistent?
>
> What are your thoughts on that?
I'd be in favor for this. We use this to put the processors into a
specific and known state until otherwise needed in a different state.
E.g. we use it to take erroring processors offline to defer swapout to a
less disruptive time or do other special performance-related tricks with
psradm.
I can see how this might be used in certain situations to stay within
terms of application licenses, too.
I might suggest psradm/psrset defaulting to persistent, but adding a
flag like -t (temporary) for one-off non-persistent use?
-Dan
|
|
0
|
|
|
|
Reply
|
Dan
|
4/12/2006 6:27:49 AM
|
|
Dan Foster wrote:
> In article <1144816749.555175.3780@u72g2000cwu.googlegroups.com>, bart.smaalders@gmail.com <bart.smaalders@gmail.com> wrote:
>
>>Perhaps psrset and psradm should be persistent?
>>
>>What are your thoughts on that?
>
>
> I'd be in favor for this. We use this to put the processors into a
> specific and known state until otherwise needed in a different state.
>
> E.g. we use it to take erroring processors offline to defer swapout to a
> less disruptive time or do other special performance-related tricks with
> psradm.
>
> I can see how this might be used in certain situations to stay within
> terms of application licenses, too.
>
> I might suggest psradm/psrset defaulting to persistent, but adding a
> flag like -t (temporary) for one-off non-persistent use?
>
> -Dan
That -t (temporary) flag would break backward compatibility. Would it not be
more sensible to have a -p (permanent) flag and keep the default as it is? That
preserves backward compatibility.
You could always write your own startup script that does this.
--
Dave K MCSE.
MCSE = Minefield Consultant and Solitaire Expert.
Please note my email address changes periodically to avoid spam.
It is always of the form: month-year@domain. Hitting reply will work
for a couple of months only. Later set it manually.
|
|
0
|
|
|
|
Reply
|
Dave
|
4/12/2006 12:38:06 PM
|
|
On 2006-04-11, Giorgos Keramidas <keramida@ceid.upatras.gr> wrote:
> Ceri Davies <ceri_usenet@submonkey.net> writes:
>> On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
>>> I have marked a number of CPUs in a machine no-intr using psradm.
>>> I can't see that this has changed anything on disk anywhere, so I am
>>> slightly worried that this change will not persist across a reboot.
>>>
>>> Will it? I am running Solaris 10, and the manuals are drawing a blank.
>>
>> The answer to that question is "no".
>>
>>> If not, is there a canonical place to put this initialisation?
>>
>> Still open!
>
> Hi Ceri :-)))
Small world :)
> I'm not sure about a "canonical" place, but an /etc/init.d script that
> runs before important stuff fire up (i.e. network interfaces) seems ok.
>
> If you feel funky, an SMF service somewhere after milestone/single-user
> would be nice too.
That's cool. So long as there is no /etc/psrset.conf or similar. Not
that there's any reason that there should be, but you never know!
Ceri
--
That must be wonderful! I don't understand it at all.
-- Moliere
|
|
0
|
|
|
|
Reply
|
Ceri
|
4/13/2006 6:03:05 PM
|
|
On Thu, 13 Apr 2006 18:03:05 GMT,
Ceri Davies <ceri_usenet@submonkey.net> wrote:
>On 2006-04-11, Giorgos Keramidas <keramida@ceid.upatras.gr> wrote:
>> I'm not sure about a "canonical" place, but an /etc/init.d
>> script that runs before important stuff fire up (i.e. network
>> interfaces) seems ok.
>>
>> If you feel funky, an SMF service somewhere after
>> milestone/single-user would be nice too.
>
> That's cool. So long as there is no /etc/psrset.conf or
> similar. Not that there's any reason that there should be, but
> you never know!
None that I know of. I usually set things up in an init.d script
of my own, saved in `/etc/init.d/processor', which is symlinked
from `/etc/rc1.d/S20psrctl' or similar :)
|
|
0
|
|
|
|
Reply
|
Giorgos
|
4/13/2006 7:11:30 PM
|
|
On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
> On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
>> I have marked a number of CPUs in a machine no-intr using psradm.
>> I can't see that this has changed anything on disk anywhere, so I am
>> slightly worried that this change will not persist across a reboot.
>>
>> Will it? I am running Solaris 10, and the manuals are drawing a blank.
>
> The answer to that question is "no".
>
>> If not, is there a canonical place to put this initialisation?
>
> Still open!
OK, I've found this.
If you enable pools:
pooladm -e
then default everything and persist the configuration to
/etc/pooladm.conf:
pooladm -x
pooladm -s
you can then ignore the big warning at the top of pooladm.conf, edit the
cpu.status of the CPUs that you want to change, and then load
pooladm.conf back in with:
pooladm -c
This will be persistent.
There is apparently some way to do it with poolcfg, but I simply cannot
understand how to modify a "component" from the manpage.
Ceri
--
That must be wonderful! I don't understand it at all.
-- Moliere
|
|
0
|
|
|
|
Reply
|
Ceri
|
4/21/2006 3:52:09 PM
|
|
On 2006-04-21, Ceri Davies <ceri_usenet@submonkey.net> wrote:
> On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
>> On 2006-04-11, Ceri Davies <ceri_usenet@submonkey.net> wrote:
>>> I have marked a number of CPUs in a machine no-intr using psradm.
>>> I can't see that this has changed anything on disk anywhere, so I am
>>> slightly worried that this change will not persist across a reboot.
>>>
>>> Will it? I am running Solaris 10, and the manuals are drawing a blank.
>>
>> The answer to that question is "no".
>>
>>> If not, is there a canonical place to put this initialisation?
>>
>> Still open!
>
> OK, I've found this.
>
> If you enable pools:
>
> pooladm -e
>
> then default everything and persist the configuration to
> /etc/pooladm.conf:
>
> pooladm -x
> pooladm -s
>
> you can then ignore the big warning at the top of pooladm.conf, edit the
> cpu.status of the CPUs that you want to change, and then load
> pooladm.conf back in with:
>
> pooladm -c
>
> This will be persistent.
>
> There is apparently some way to do it with poolcfg, but I simply cannot
> understand how to modify a "component" from the manpage.
That process is actually somewhat more stupid than it needs to be, but
it wasn't the only stupid thing I did yesterday afternoon so I'm just
putting it down to Friday.
The easy way to do it is to make your changes with psradm:
psradm -i 1-3 5-7 9-11 13-15 17-19 21-23 25-27 29-31
Then, enable pools and just write out the config to disk:
pooladm -e
pooladm -s
Ceri
--
That must be wonderful! I don't understand it at all.
-- Moliere
|
|
0
|
|
|
|
Reply
|
Ceri
|
4/22/2006 10:08:41 AM
|
|
|
9 Replies
576 Views
(page loaded in 0.112 seconds)
Similiar Articles: Are changes made with psradm persistent across reboot? - comp.unix ...I have marked a number of CPUs in a machine no-intr using psradm. I can't see that this has changed anything on disk anywhere, so I am slightly worrie... Which drives support SCSI-3 Peristent Reservations? - comp.periphs ...... ROM drive in a SCSI system - comp.periphs.scsi Which drives support SCSI-3 Peristent Reservations? - comp.periphs ... Are changes made with psradm persistent across reboot ... number of CPUs - comp.unix.solarisAre changes made with psradm persistent across reboot? - comp.unix ... I have marked a number of CPUs in a machine no-intr using psradm. I can't see that this has changed ... ATAPI CD-ROM drive in a SCSI system - comp.periphs.scsiAre changes made with psradm persistent across reboot ... Remedy tech support told me back in late ... ATAPI CD-ROM drive in a SCSI system - comp.periphs.scsi ... ntpd, boot time, and hot plugging - comp.protocols.time.ntp ...One of the things that slows down the boot pr... ... that you can get this down to about seven seconds across ... Or, you could monitor syslog output, or look for changes ... [comp.publish.cdrom] CD-Recordable FAQ, Part 1/4 - comp.publish ...Archive-name: cdrom/cd-recordable/part1 Posting-Frequency: monthly Last-modified: 2008/10/09 Version: 2.71 Send corrections and updates to And... Are changes made with psradm persistent across reboot? - comp.unix ...I have marked a number of CPUs in a machine no-intr using psradm. I can't see that this has changed anything on disk anywhere, so I am slightly worrie... Solaris General - - Apache, Linux, MySQL, Websphere, IIS, Sendmail ...Re: Are changes made with psradm persistent across reboot? Posts(4) April 27, 2006, 7:59 am: sunsolve patch access broken: Posts(2) April 27, 2006, 7:59 am 7/24/2012 7:29:26 PM
|