Re: My woes continue

From: <kasperle_at_cs.tu-berlin.de>
Date: 1997/07/10
Message-ID: <33C4CE1E.937_at_cs.tu-berlin.de>#1/1


Joost Ouwerkerk wrote:
>
> To all DBA's:
> My DBA is making my job rather difficult, but I'm having trouble
> exposing some of his incompetencies to my superiors... Should any of
> the following statements be true for a well-tuned Oracle 7.3.2
> database server running on an IBM AIX and serving an organization of
> 50 users, about 15 of whom are data entry staff, and about 10 of whom
> run regular queries and reports:
>
> 1) A complex query can 'crash' the database and cause corruption of
> data.

[Quoted] Yes it can (done it myself). If a query is _that_ complex, the server can get _that_ overwhelmed, that it just isn't able anymore to cry for help.
Mine came down with a core dump.
No corruption, though. (can't remember exactly. it's an enginering system we quite often shutdown and startup, sometimes even for power failures, so corruption isn't anything horrible to us :-))

> 2) Running Personal Oracle on a client (user) machine can 'crash' the
> database and cause corruption of data
> 3) Running Developer/2000 on a client machine can 'crash' the database
[Quoted] > and cause corruption of data.

I don't think, that it makes any difference which tool starts the difficult query. In my case, I put it in a stored procedure. Anyone who can invoke this, might crash.

> My suspicion is that if any of these are true, there is a server
> tuning problem. My DBA is suggesting no queries be run while data
> entry is occuring (during business hours) and is threatening to remove
> Personal Oracle and Dev/2000 from my computer because of the damage
> I'm supposedly causing. Am I going crazy?

He's right. Engineers shouldn't try risky thinks in working hours on production systems. Go to the off hours or make up a test system. You should know, whether things get risky (at least after the first crash).

The easy solution to my problem was, to get a _big_ rollback segment online. And to use it for the query.
After a while, I learned, that tuning the DB (extents of some log-tables) would do the trick and even better. After another while, I learned that tuning the query would even do the trick, too.

So don't fight the DBA, try to fix it yourself.

regards,
Kathrin Received on Thu Jul 10 1997 - 00:00:00 CEST

Original text of this message