[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: real world examples
I've put a small collection of logs together:
http://www.exratio.com/test.zip. The test code is in there too -- nothing
fancy. Just spawns eight threads and does some random stuff like creating
and deleting objects in the generic data layer package I'm writing
(www.twilightminds.com/jdffull.html).
Annoyingly, it's one of those problems in which turning debugging on and off
changes what actually happens. Oh well. In every case, the server freezes
up; this happens in every run of the test code, even with randomizing of the
order in which operations are performed by the threads. If debugging is on,
the JVM crashes with a stack overflow. If debugging is off, things just
freeze.
"Exception: org.ozoneDB.ObjectNotFoundExc: No such object." is fine -- I
expect that. They occur here and there in the test logs due to the
hacked-together way I'm selecting objects for deletion. Not a problem.
At some points in time, Ozone is throwing a NullPointerException.
Originally, I was thinking that that this precipitated a race condition
fairly quickly, but it looks like I may be wrong about that. This occurs in
org.ozoneDB.core.wizardStore.WizardStore.containerForID(WizardStore.java:377
)
I looked at this and surrounding code, and it makes no sense to me at all
that this would come up with a NullPointerException in the first place.
Looks sound.
So it looks like deadlocking and race conditions occur every time; if those
more familiar with Ozone than I could take a look and make some educated
guesses? I've been looking at the internals of Ozone that deal with this
stuff and haven't made much headway in comprehension.
Reason
-----Original Message-----
From: ozone-users-owner@ozone-db.org
[mailto:ozone-users-owner@ozone-db.org]On Behalf Of Falko Braeutigam
Sent: Monday, June 18, 2001 11:39 AM
To: Reason; ozone-users@ozone-db.org
Subject: RE: real world examples
On Mon, 18 Jun 2001, Reason wrote:
> > > Now I am planning a C/S remote control system and I want some
> > more reference material to see
> > > the prossiblity of using Ozone in the background.
> >
> > As always, if ozone suits your needs highly depends on your particular
> > requirements. The best way to find out is to make a (very)
> > limited prototype
> > and just test it.
> >
> > Anyway, some reference projects would be fine anyhow. So please guys,
help
> > us by giving back your real-world experiences. This would help
> > newbies a lot to find out about ozone, gets us more users, gets us
better
> > visibility, gets us ahead. Thanks ;)
>
> My real-world experience:
Thanks:)
>
> I'm having considerable difficulty producing a solution for a generic data
> layer with Ozone that is stable under multithreading -- external
> transactions kill me. Or rather, they crash the server for even simple
> applications providing that the threads are following one another closely
> enough.
Hmmm... so there must be some kind of race condition inside the
ExternalDatabase code.
What does this mean "crash the server"? Do you have a stack trace or
something?
I never, under no circumstances, saw an ozone server actually "crash"
(actually
stop execution), with the exception of JVM crashes.
Sorry, if you already provided this information.
>
> Admittedly, this may be the sort of project that isn't completely suited
to
> OODBMS -- i.e. some aspects of business logic for an app built in the
> generic data layer are not going to make it into the database server --
but
> I can code the complex and potentially time-consuming low-level stuff
(like
> non-trivial cross-referencing and searching in large collections of
objects)
> to run inside the database server. So I view this as a worthwhile project.
> Provided of course I can get it to work :)
So can we add it to our reference page? Together with some sentences about
the
role of ozone in the project, status, plans and problems regarding ozone?
Falko
>
> http://www.twilightminds.com/jdffull.html
>
> (OTOH, I haven't tested Ozone's implicit transactions under high loads for
a
> while, so I'm not sure the problem is restricted to external
transactions).
>
> Reason
> http://www.exratio.com/
--
______________________________________________________________________
Falko Braeutigam mailto:falko@smb-tec.com
SMB GmbH http://www.smb-tec.com