Subject Re: Firebird Restarting part 2
Author Michael Vilhelmsen
--- In firebird-support@yahoogroups.com, "Alan McDonald" <alan@...> wrote:
>
> > Yesterday was the first day with all users connected and the sweep
> > disabled.
> >
> > The GAB became as large as 200K, but noone noticed.
> > When everybody disconnected at nigth it shrank down to 1.
> >
> > The sweep ran in the middle of the nigth and took 25 minutes.
> >
> > The memory use was significantly smaller at the time where it has
> > prev. been a problem.
> >
> > So all in all (based on 1 day of observation ;-) ) I think the sweep
> > did take a lot of res. and memory.
> >
> >
> > Another thing I tried (on a diff. DB, a smaller customer) was forcing
> > a disconnect from the DB and a connect to the DB every once in a
while.
> > This customer now has a gab between 50 and 400. Usually it was
> > somewhere between 500 and 6000.
> >
> > I think I'll try that on this big customer DB as well.
> >
> > Anywas with any comments on this ?
> >
> > Michael
>
> forcing the disconnect..? how? shutdown? or asking them to close?
> still sounds like a long running transaction to me. Maybe something
you have
> had all the time and still haven't found it in your code.


I do believe that we somewhere has a long running query.
Or that one of the users does something every morning without
comitting it.
I'll have to look into this....

The 3 appl. we have - 2 of them are made, that once they are started
they have no open querys of any kind. Its only when they do something,
that the open querys, do insert and updates etc.

Does to I'm fairly sure doesn't have the problem.


The last one wasn't originally written by me.
I have been through almost every part of the program. At least Every
part that is being used very often.

Just a few days ago I changed they way to program works.
Whenever they exit a menu, they will close every open query, commit
every active transaction, disconnect and not reconnect until they
enter a new menu.
The above solution has made the gab significantly smaller.

Because of this I think that somewhere here, the long running query is
to be found.
When we switch this appl. from IBX to FIB I'll have to look intom
every part of the program, to see that it still does was is expected.
Here I'll also look at the transactions, and see if something can be
done another way.


But for now - It runs better than it has for a long time.
And this is quite nice....

Thx for every input you have made so far.
You have been very helpfull.

Michael
> Alan
>