onmode -BC

I configured and initialized a brand new installation of 9.4 on a new machine, today. I initialized with a rootdbs of about 62GB, and a tempdbs of about 8 GB. Everything looks OK with respect to the large chunks, but I realized afterwards that I never executed onmode -BC (either 1 or 2). I guess I misunderstood about 'onmode -BC', right? I mean, it just worked with large chunks, right "out of the box". DG -- David Grove - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - "I think not," said Descartes, and disappeared.

David E. Grove wrote: > I configured and initialized a brand new installation of 9.4 on a new > machine, today. I initialized with a rootdbs of about 62GB, and a > tempdbs of about 8 GB. > > Everything looks OK with respect to the large chunks, but I realized > afterwards that I never executed onmode -BC (either 1 or 2). > > I guess I misunderstood about 'onmode -BC', right? I mean, it just > worked with large chunks, right "out of the box". I tried for large chunks on Solaris recently before I knew about -BC and it refused. What platform? Are there some with it defaulted on? -- I drink therefore I, umm......

David E. Grove wrote: oh - maybe you put very large numbers into the config file for rootdbs, and it automagically knew it needed to effect a -BC?

----- Original Message ----- From: "David E. Grove" <david_grove@correct.state.ak.us> Newsgroups: comp.databases.informix Sent: Friday, September 17, 2004 1:23 AM Subject: onmode -BC > I configured and initialized a brand new installation of 9.4 on a new > machine, today. I initialized with a rootdbs of about 62GB, and a tempdbs > of about 8 GB. > > Everything looks OK with respect to the large chunks, but I realized > afterwards that I never executed onmode -BC (either 1 or 2). > > I guess I misunderstood about 'onmode -BC', right? I mean, it just worked > with large chunks, right "out of the box". As Andrew tried to say, I think, it doesn't work straight out the box for an existing database server: you have to take a conscious decision to enable the large chunks . It seems that it is just a conversion feature: ie you can create a new one straight off. What does your new instance say at the bottom of onstat -d?

Yep --- can't very well not have BC2 if you have a root DBS greater than 2GB. This only happens with an initial install and when big chunk features are required during oninit -iy. "Andrew Hamm" <ahamm@mail.com> wrote in message news:2qupt5F13tbrvU1@uni-berlin.de... > David E. Grove wrote: > > oh - maybe you put very large numbers into the config file for rootdbs, > and it automagically knew it needed to effect a -BC? > >

If you run oninit -iy with an initial chunk greater than 2GB, then BC2 mode is automatically enabled. "David E. Grove" <david_grove@correct.state.ak.us> wrote in message news:10kkbjus21n6e25@corp.supernews.com... > I configured and initialized a brand new installation of 9.4 on a new > machine, today. I initialized with a rootdbs of about 62GB, and a tempdbs > of about 8 GB. > > Everything looks OK with respect to the large chunks, but I realized > afterwards that I never executed onmode -BC (either 1 or 2). > > I guess I misunderstood about 'onmode -BC', right? I mean, it just worked > with large chunks, right "out of the box". > > DG > > > > > > -- > David Grove > > - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - > "I think not," said Descartes, and disappeared. > >

Neil Truby wrote: > > As Andrew tried to say, I think, it doesn't work straight out the box > for an existing database server: you have to take a conscious > decision to enable the large chunks . It seems that it is just a > conversion feature: ie you can create a new one straight off. What > does your new instance say at the bottom of onstat -d? Is that what I was trying to say? thanks for explaining :-) Actually I was muddling up a thought that it does an implicit -BC if is sees a very large root allocation when initialised, because for me, this mode was disabled when i started with a small rootdbs. So that sorta implies to me it will assume -BC IF you have a large root allocation in the config when initialized. It does sound better that it would be enabled for all new engines, but it didn't happen for me on Solaris.

Madison Pruet wrote: > If you run oninit -iy with an initial chunk greater than 2GB, then > BC2 mode is automatically enabled. What's the reason for not enabling it for all new initialisations? Neil's suggestion that all new inits should get BC2 sounds more forward-looking... Of course all recoveries should do exactly what the tape tells them.

"Andrew Hamm" <ahamm@mail.com> wrote in message news:2qv7c6F12r2uoU1@uni-berlin.de... > Neil Truby wrote: > > > > As Andrew tried to say, I think, it doesn't work straight out the box > > for an existing database server: you have to take a conscious > > decision to enable the large chunks . It seems that it is just a > > conversion feature: ie you can create a new one straight off. What > > does your new instance say at the bottom of onstat -d? > > Is that what I was trying to say? thanks for explaining :-) Actually I was > muddling up a thought that it does an implicit -BC if is sees a very large > root allocation when initialised, because for me, this mode was disabled > when i started with a small rootdbs. So that sorta implies to me it will > assume -BC IF you have a large root allocation in the config when > initialized. It does sound better that it would be enabled for all new > engines, but it didn't happen for me on Solaris. It was a case of 'fear factor'. > >

"Andrew Hamm" <ahamm@mail.com> wrote in message news:2qv7f5F1441elU1@uni-berlin.de... > Madison Pruet wrote: > > If you run oninit -iy with an initial chunk greater than 2GB, then > > BC2 mode is automatically enabled. > > What's the reason for not enabling it for all new initialisations? Neil's > suggestion that all new inits should get BC2 sounds more > forward-looking... > > Of course all recoveries should do exactly what the tape tells them. Because we needed to allow the test suites to still function with old format disks so that we could run QA with what the existing customer base had installed. > >

Madison Pruet wrote: > > It was a case of 'fear factor'. Fair call. 9.50 perhaps?

"Andrew Hamm" <ahamm@mail.com> wrote in message news:2qv7f5F1441elU1@uni-berlin.de... > Madison Pruet wrote: > > If you run oninit -iy with an initial chunk greater than 2GB, then > > BC2 mode is automatically enabled. > > What's the reason for not enabling it for all new initialisations? Neil's > suggestion that all new inits should get BC2 sounds more > forward-looking... We needed a way to create a server in the old format so that we could still run QA in the same format that the customers are in as well as to be able to use the new format. > > Of course all recoveries should do exactly what the tape tells them. > >

Andrew Hamm wrote: > David E. Grove wrote: > > oh - maybe you put very large numbers into the config file for rootdbs, > and it automagically knew it needed to effect a -BC? Yes. If the first chunk of the rootdbs won't fit into 2 GB (because you said its bigger than that), you don't have to futz with 'onmode -BC'. If you create the (first chunk of the) rootdbs smaller than 2GB, you do. -- Jonathan Leffler #include <disclaimer.h> Email: jleffler@earthlink.net, jleffler@us.ibm.com Guardian of DBD::Informix v2003.04 -- http://dbi.perl.org/

Madison Pruet wrote: > "Andrew Hamm" <ahamm@mail.com> wrote in message > news:2qv7c6F12r2uoU1@uni-berlin.de... > >>Neil Truby wrote: >> >>>As Andrew tried to say, I think, it doesn't work straight out the box >>>for an existing database server: you have to take a conscious >>>decision to enable the large chunks . It seems that it is just a >>>conversion feature: ie you can create a new one straight off. What >>>does your new instance say at the bottom of onstat -d? >> >>Is that what I was trying to say? thanks for explaining :-) Actually I was >>muddling up a thought that it does an implicit -BC if is sees a very large >>root allocation when initialised, because for me, this mode was disabled >>when i started with a small rootdbs. So that sorta implies to me it will >>assume -BC IF you have a large root allocation in the config when >>initialized. It does sound better that it would be enabled for all new >>engines, but it didn't happen for me on Solaris. > > It was a case of 'fear factor'. Or insufficient fear factor - I didn't rant loud enough in time because it didn't occur to me that they wouldn't do it automatically so I didn't find out about it in time to make them quake in their boots and fix it before it got released. -- Jonathan Leffler #include <disclaimer.h> Email: jleffler@earthlink.net, jleffler@us.ibm.com Guardian of DBD::Informix v2003.04 -- http://dbi.perl.org/