Sorry for the delay.

You could add a flag in TaskStorageManager which will skip all the function
call inside the 'init' method if the directory obtained by
'TaskStorageManager.getStorePartitionDir' already exists. Again - super
hacky - but should make your life easier.

On Sat, Jan 17, 2015 at 11:42 AM, Roger Hoover <[email protected]>
wrote:

> Chinmay,
>
> Thanks for the reply.  I realize that a full-blown solution would probably
> require significant design and implementation but I wonder if there could
> be an "unsafe" development mode.  I'm trying to benchmark a job but every
> time I make a change and restart it, I have to wait many minutes to reload
> all the state.
>
> On Fri, Jan 16, 2015 at 9:46 PM, Chinmay Soman <[email protected]>
> wrote:
>
> > Unfortunately, not yet :(  Currently, Samza will destroy the local state
> > and rebuild it (if the changelog is enabled) even if it is on the same
> > machine.
> >
> > Having sticky data would be awesome, but we would need some way to verify
> > that the data is sane (and not corrupted).
> >
> > On Fri, Jan 16, 2015 at 9:39 PM, Roger Hoover <[email protected]>
> > wrote:
> >
> > > Hi,
> > >
> > > If a Samza job is restarted on the same machine where it just ran and
> all
> > > the partitioning is the same, is there a way for it to reuse it's
> > existing
> > > local state?
> > >
> > > Especially for development, this would be really handy.
> > >
> > > Thanks,
> > >
> > > Roger
> > >
> >
> >
> >
> > --
> > Thanks and regards
> >
> > Chinmay Soman
> >
>



-- 
Thanks and regards

Chinmay Soman

Reply via email to