Re: Roll-back program questions

[Alan Campin <alan0307d@xxxxxxxxx>]

Have you considered using my Make tool (COMPILE) at
www.think400.dk/downloads.htm. You can put your instructions for how to
create the objects in the header and have your change management run the
COMPILE command to create the object.


On Tue, May 21, 2013 at 11:03 AM, Koester, Michael
<mkoester@xxxxxxxxxxxxx>wrote:

> We use a home-grown (circa 1980's) change management system that does a
> pretty good job of archiving source, but was not designed to handle the
> various ILE objects I'm now using, so I've been augmenting our turn-over
> process with a CL program to create *PGM, *BNDDIR, *SRVPRM objects in the
> desired order with the desired optional parameters, etc., to simplify the
> SysAdmin's life.  I would like to add some code to these "implementer
> programs" to facilitate an easy/speedy recovery if I should ever screw it
> up and need to roll back my changes.  We do not use journaling, and
> recovery from tape might be slower than acceptable.
>
> We develop and test on the same box, using test database objects and
> production application objects, except where our test library (higher in
> our test environment *LIBL) may have a new-and-improved version.  Been at
> 7.1 for a long time now.
>
> A typical turnover these days may involve:
>         OPM stuff (created from PDM option 14),
>         binder source updates,
>         *PGMs created from source,
>         *MODULEs created from source,
>         *PGMs created from *MODULE,
>         *SRVPGMs created/updated with CRTSRVPGM or UPDSRVPGM commands,
>         *BNDDIRs created/updated with CRTBNDDIR and ADDBNDDIRE commands
>
> We don't mess with *LIB objects, and updates to websites are handled
> somewhat differently.
> The database may get a new file or two (from DDS source), but we aren't
> using DDL, triggers, constraints, views, etc.
>
> In the event a roll-back was required, users of the effected applications
> would be asked to stay out until the roll-back was complete, so operating a
> partially restored application is not an issue.
>
> My thought was that I would add CRTDUPOBJ statements for each object being
> changed to the "implementer program".  I would then prepare a recovery CL
> that would be able to dltobj each of the new/changed pieces and then rnmobj
> the copies to restore everything to their original places.
>
> My questions:  Are there object types that would not be easily recoverable
> in this manner?
> What objects might be especially sensitive to the existence of others, and
> thus require that sequence is critical? Or does order not matter, so long
> as all the original objects are renamed to their original places before
> they are accessed?
> I'm guessing that binder source is only used with CRTSRVPGM, so that could
> be restored to original after the objects are in place and users are back
> on-line? In an urgent situation, I'd envision getting the objects back in
> place, getting the users on-line and happy, and then getting all the source
> (of all types) back where it belongs.
>
> I appreciate the advice,
>
> Michael Koester
>
> --
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.