Hi there,
We have a production problem at the moment as follows: A WAS Java is running on MaxDB (7.5.00.32; I know, it is old), had no issues for years.
Now, since 12 days, the system halts every morning, due to the DB not answering any more. On the DB side, normally (after a WAS / DB restart), the DATA area is filled up to 79%. We observed now that during the issue, this goes up to > 95%. So I expect this coming from results, as this space gets free after a restart immediately.
We also had a bad design for the DATA area (one DATA file with 15GB). We will change this within the next hours to 5 * 5GB DATA files, removing the 15GB DATA file afterwards. That should (a) offer more parallelism (we see many IOWaits and "IOWait(R)(041): b13get_node: await read :-(" during the critical situation), and (b) offer enough space.
If this succeeds (any additional thoughts welcome), I would like to analyse which statements lead to big results, as probably this is what has changed 12 days ago. Question: How to check this?
Thanks in advance
Detlev