Re: [Sqlrelay-discussion] ping() in Java

[David M. <dav...@fi...>]

On Fri, 2005-12-16 at 23:55 +0100, Maciej Wisniowski wrote:
> Hi!
> 
> Thanks for the answer.
> 
>  > I just ran some tests and I got the same behavior from python and java.
>  > What version of SQL Relay are you using? Maybe there's a bug in that
>  > version.
> I can't check my configuration now but AFAIR it is 0.36 (I'll
> check this in monday).
> 
>  > Give the test programs below a try and see if you don't get the same
>  > results as the log below shows.
> I'll try this. The problem is that I'm not the database admin and so far
> I was only able to check this behaviour once a day...
> Next week I hope we'll have time to do some work with this java app
> and create another oracle instance for testing only.
> 
>  > If SQL Relay lost contact with the database during the backup and no
>  > other queries had been run, then I'd expect the ping to return false,
>  > and then the next query to cause SQL Relay to reconnect, run the query
>  > and return rowcount of 1.
> This is strange for me. I expect ping to reconnect and return
> true in this situation. Now it seems that ping is not checking database
> but rather listener's status...? What for is this ping then?
> What logic is in that it returns 'false' even if it is possible to get
> data from database?

Oops, I just realized there was a question here while moving messages
around :)  Sorry I didn't answer it earlier....


ping tells you whether the connection between SQL Relay and the database
is ok or not.  I implemented it that way because someone requested it,
but I don't remember who, it was a long time ago :)

I think back then, SQL Relay didn't automatically recover when a
database connection went down.  So, it might have been more useful then
than it is now.

> 
> 
>  > If you have 3 sqlr-connection daemons running, you'd see that behavior 3
>  > times. After all 3 sqlr-connections have re-established their database
>  > conections, ping should start returning true.
> I will play with this.
> 
>  > The behavior your java program exhibits is exactly what I'd expect.
>  > If the python program exhibits different behavior, then maybe there's a
>  > bug there. Although ping is implemented identically for both API's.
>  > They just wrap the C++ function.
>  > Does the python program maybe run against a different database, where
>  > another program may have run queries against the database since the
>  > backup?
> I'm not sure. Something like that could happened. It's possible that some
> queries were done, maybe from java... I did not know the rule of 3
> sqlr-connections ;) so maybe my interpretation was wrong and python
> is working like java... I'll check this next week.
> 
>