Thanks for the update. Is the XM buffer dynamically sized by MPE as
required by demand? Or is it still a fixed (albeit user-configurable)
size that will cause XM to abort the transaction if it fills up?

Steve


> The problem of small XM buffer has been addressed. Now there
> is a large
> and configurable XM buffer, and is capable of supporting large
> transactions. This solution has gone out with MPE 6.5. I am also
> interested in knowing the specifics of  any other dynamic rollback
> problem.
>
> Regards,
> Vikram
>
> Steve Dirickson wrote:
>
> > > IMO HP appears to treat the dynamic rollback problem more like
> > > a DBMS limitation than a bug.
> >
> > Pardon my ignorance: what "dynamic rollback problem" are we
talking
> > about? The XM buffer overflow is a defect in MPE, not in
> the DBMS, so
> > I assume this refers to something else?
> >
> > Steve
>