[rt-users] [SLA] Custom (and very imperfect) solution for parking stalled tickets

[Robert Wysocki]

Dnia 2012-06-12, wto o godzinie 18:24 +0400, Ruslan Zakirov pisze:
> 
> Sorry. Didn't notice that.
> 
> I'm not sure why you need additional data storage. 

Additional storage in form of CF's in my solution is merely a product of
the lack of time. I needed it to work and the simplest way was then for
me to code it like this.
Now I see some other options; I belive that if I move the scrip-part of
the code to module code I won't have the need for additional storage any
more.

> I see putting Due
> date on hold in the following way:
> 
> 1) When status is changed from initial/active to some that marked as
> "on hold", we just unset Due date. We can not keep old value in the
> Due field as it will mess sorting of tickets.
> 
> 2) When status is changed from "on hold" to any active, we
> re-calculate Due date.

That sounds good.

> Re-calculation is hard to make sane. RT out of the box opens tickets
> on replies, so reply and activation from "on hold" events match. 

In our case we decided that in the event of re-activating ticket after
it was parked for some time we change it's status to "open". So we
really don't care about what other scrips do to tickets' status - our
logic goes the other way.

> This
> case is very simple, we treat it as any other reply.

We shoudn't change status to "open" if reply wasn't from one of the
requestors - ticket should stay parked.

> It becomes questionable when people disable "auto open" on some
> replies. There are several interesting timelines that may happen:
> 
> 1) reply ... on hold ... no replies ... activation
> 2) reply ... on hold ... reply(ies) ... activation
> 
> In first case due date can be calculated from reply plus time ticket
> was on hold. Second case is harder and I'm still not sure how to treat
> it.

In our logic there's no place for ticket activation when there was no
reply from one of the requestors.
In other words if someone will activate a parked ticket "by hand"
recalculation won't happen.

Best regards,

-- 
Robert Wysocki
administrator systemów linuksowych
Contium S.A., http://www.contium.pl