[Info-ingres] [Info-Ingres] Ingres 9.2 to 10.1 Migration

[nikosv]

are there any steps that should be taken for switching application code from using ANSI isolation to MVCC?

On Friday, August 8, 2014 2:24:15 PM UTC+3, Sørensen. Henrik Georg HGO wrote:
> We did the migration 1 year ago from Ingres 9.2.3 to 10.1.0, and as remember only some issues,
> 
> that should be solved by the present patches, except may be this one, related to the Database.
> 
>  
> 
> If using portioned tables there was an issue, so before migrating the database we've did a un-partitioning of these
> 
> Tables, migrated and did a re-partioning of the tables.
> 
> (No problem if you have scripts for the structure, and sorry we didn't report it to support at that time).
> 
>  
> 
> In our case (and my knowledge of past migrations of Ingres versions) we also removed all of our own DB-rules,
> 
> DB-Procedures and Views, as well as Grants. (Jobs and Scripts as well to reload them).
> 
> This mainly as it often seems faster to do this, than let the Migration do this, and less chances of failures anyway.
> 
>  
> 
> But if you're moving platform with unload/reloading of database this shouldn't be a big issue anyway.
> 
>  
> 
> During the upgrade of Ingres versions we only did the Ingres internal migration for iidbdb and imadb, and waited with
> 
> Our own database until after the installations we're patch up to the latest at that time.
> 
>  
> 
> At the moment only one open Issue on Ingres 10s (10.1 Linux 64bit), so don't use a where criteria with 1 = 0
> 
> On a update or select as you properly would affect all rows instead of none (see previously mail to newsgroup still not solved!).
> 
>  
> 
> For your programs remember that Ingres 10 and upwards now support longer table and fieldnames, so if your
> 
> Programs support names for DB-objects in variables remember to raise the typical length from 32 to 64/256,
> 
> And so.
> 
>  
> 
> Migrations between Unix-versions could as well give some differences in C-code, Shell-scripts but that is not directly
> 
> Releated to Ingres, and would also depend on coding styles (only Ingres abf-issue was about removing ; on lines
> 
> With #include <inc/mydefs.h>; witch is more cpp-related than Ingres).
> 
>  
> 
> Had a lot of different issues during past migration a couple of years ago from TRU64 (axp.osf) to Linux,
> 
> about old-style C-code, Environment/permissions in Shell-scripts, Different flags and so on.
> 
>  
> 
> 
> 
> 
> 
> Kind regards