oncheck -me
A user spotted an undocumented-looking trick in an IBM performance tutorial PDF: setting RASHELP=1 and running 'oncheck -me <partnum> <extents>' to merge/reorg a table's extents dynamically. He found it worked only on one platform/version he tested (Alpha, IDS 7.30) and asked who uses it and where it's enabled. Madison Pruet (IBM) replied that this is an internal RAS feature intended only for tech support, largely untested, unsupported, probably not recoverable, and enabled on only a few ports; its appearance in the public tutorial was a mistake, and the documents might be pulled. Art Kagel reported it working well and fast on 7.31UD4/DGUX and urged IBM to QA and document it properly. No fix or official support was promised.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
hi all
In the last IIUG Insider #38 was link to
http://www-3.ibm.com/software/data/support/pdm/idstutorials.html
In the "Performance Problem Determination" (
ftp://ftp.software.ibm.com/software/data/support/pdm/idstut7.pdf )
i've found one very interesting IDS feature - dynamic table extents merging:
1) export RASHELP=1
2) oncheck -me <partnum> <target extents number>
I try to test this functionality on several platforms && IDS versions.
The only working case was on Digital Alpha Server, IDS 7.30
So my questions are :
- Does anybody use oncheck -me ? Ain't there any troubles with it ?
- What platforms/IDS versions have this functionality enabled ?
Thanks.
> 1) export RASHELP=1
> 2) oncheck -me <partnum> <target extents number>
>
> I try to test this functionality on several platforms && IDS versions.
>
> The only working case was on Digital Alpha Server, IDS 7.30
>
> So my questions are :
> - Does anybody use oncheck -me ? Ain't there any troubles with it ?
Does anyone use it - I hope not.
Are there problems with is - probably
This functionality was intended to be used only by tech support, not as a
generalized user functionality. It has not received any serious testing
that I know of. I don't know why it was documented at all since it is not
supported.
> - What platforms/IDS versions have this functionality enabled ?
only a few...
M.Pruet
>
> Thanks.
On Tue, 21 Oct 2003 13:29:29 -0400, Madison Pruet wrote:
Tried it today on a development instance IDS 7.31UD4 on DGUX to reorg a 540000
page, 5 extent, table with over 2million rows and eleven detached indexes of
varying key width. Worked fine, seemed to take no logical log space and only one
lock and was at least as fast as ALTER FRAGMENT ON TABLE ... INIT IN ...;
Considering how often we (ie us DBAs) do table reorgs, this would be a great
tool to QA and document, Madison!
Art S. Kagel
>> 1) export RASHELP=1
>> 2) oncheck -me <partnum> <target extents number>
>>
>> I try to test this functionality on several platforms && IDS versions.
>>
>> The only working case was on Digital Alpha Server, IDS 7.30
>>
>> So my questions are :
>> - Does anybody use oncheck -me ? Ain't there any troubles with it ?
>
> Does anyone use it - I hope not.
> Are there problems with is - probably
>
> This functionality was intended to be used only by tech support, not as a
> generalized user functionality. It has not received any serious testing that
> I know of. I don't know why it was documented at all since it is not
> supported.
>
>> - What platforms/IDS versions have this functionality enabled ?
>
> only a few...
>
> M.Pruet
>
>
>
>> Thanks.
Yep - uses very little logs.
But I don't think that it is recoverable.
I agree that the feature is much needed, but a feature needs to be tested
and stable before it is documented.
The RAS features (RASHELP) were not supposed to be used except by tech
support.
M.P.
"Art S. Kagel" <kagel@bloomberg.net> wrote in message
news:pan.2003.10.21.17.10.44.248280.10594@bloomberg.net...
> On Tue, 21 Oct 2003 13:29:29 -0400, Madison Pruet wrote:
>
> Tried it today on a development instance IDS 7.31UD4 on DGUX to reorg a
540000
> page, 5 extent, table with over 2million rows and eleven detached indexes
of
> varying key width. Worked fine, seemed to take no logical log space and
only one
> lock and was at least as fast as ALTER FRAGMENT ON TABLE ... INIT IN ...;
>
> Considering how often we (ie us DBAs) do table reorgs, this would be a
great
> tool to QA and document, Madison!
>
> Art S. Kagel
>
> >> 1) export RASHELP=1
> >> 2) oncheck -me <partnum> <target extents number>
> >>
> >> I try to test this functionality on several platforms && IDS versions.
> >>
> >> The only working case was on Digital Alpha Server, IDS 7.30
> >>
> >> So my questions are :
> >> - Does anybody use oncheck -me ? Ain't there any troubles with it ?
> >
> > Does anyone use it - I hope not.
> > Are there problems with is - probably
> >
> > This functionality was intended to be used only by tech support, not as
a
> > generalized user functionality. It has not received any serious testing
that
> > I know of. I don't know why it was documented at all since it is not
> > supported.
> >
> >> - What platforms/IDS versions have this functionality enabled ?
> >
> > only a few...
> >
> > M.Pruet
> >
> >
> >
> >> Thanks.
Madison Pruet wrote: > Yep - uses very little logs. > But I don't think that it is recoverable. > > I agree that the feature is much needed, but a feature needs to be > tested and stable before it is documented. > > The RAS features (RASHELP) were not supposed to be used except by tech > support. Then somebody made a boo-boo letting that one slip out to a training document?
Yep "Andrew Hamm" <ahamm@mail.com> wrote in message news:bn4i2k$s4ogd$1@ID-79573.news.uni-berlin.de... > Madison Pruet wrote: > > Yep - uses very little logs. > > But I don't think that it is recoverable. > > > > I agree that the feature is much needed, but a feature needs to be > > tested and stable before it is documented. > > > > The RAS features (RASHELP) were not supposed to be used except by tech > > support. > > Then somebody made a boo-boo letting that one slip out to a training > document? > >
Andrew Hamm wrote: > Madison Pruet wrote: > >>Yep - uses very little logs. >>But I don't think that it is recoverable. >> >>I agree that the feature is much needed, but a feature needs to be >>tested and stable before it is documented. >> >>The RAS features (RASHELP) were not supposed to be used except by tech >>support. > > > Then somebody made a boo-boo letting that one slip out to a training > document? That one for sure - I'm not wholly convinced that any of them should have been released. I didn't notice that it was on the externally accessible web site at first; I assumed the tutorials were on the internal web only until enlightened by Madison. They may be removed - in fact, someone reported having problems accessing them and it may be because they are no longer there.... -- Jonathan Leffler #include <disclaimer.h> Email: jleffler@earthlink.net, jleffler@us.ibm.com Guardian of DBD::Informix v2003.04 -- http://dbi.perl.org/
Isn't RASHELP discussed as part of the internal course? Andrew Hamm wrote: > > Madison Pruet wrote: > > Yep - uses very little logs. > > But I don't think that it is recoverable. > > > > I agree that the feature is much needed, but a feature needs to be > > tested and stable before it is documented. > > > > The RAS features (RASHELP) were not supposed to be used except by tech > > support. > > Then somebody made a boo-boo letting that one slip out to a training > document? -- Paul Watson # Oninit Ltd # Growing old is mandatory Tel: +44 1436 672201 # Growing up is optional Fax: +44 1436 678693 # Mob: +44 7818 003457 # www.oninit.com #