BI Growth

[SSuhaib]

Hi,

We have a small database of 1 GB and we moved it out to another server. Since then the bi file grew to 3 gb and the DB crashed. Then we restored DB and applied ai files and added 2 1gb fixed and 1 variable in the st file. After 2 days the variable extent hit 2gb limit and the overall bi size now is 4 gb.

The developer say his program is correct since it was the same program running in the old server without any problems. Also there is no modification of schema.

Both the times we tried creating another st file with extra extents but the failed.

Currently we are using Progress 9.1D but we are in the process of moving to 10.1C and OS is redhat.

Thanks in advance.

 

[tamhas]

BI growth like thsi is generally not the fault of one bad program, but rather of a long running transaction which keeps BI clusters from being reused. Either way, it *is* the program(s) and how they are being used, not some OS thing.

 

[TomBascom]

I would be suspicious that something, perhaps a batch program, that used to work in the old environment is not working properly in the new.

You should be able to discover the source of the problem by simply looking in ProTop (or PROMON) for the oldest active transaction. The usr# shown is the culprit. Once you know which session is causing the problem you then have to figure out what that session is doing. That can be more difficult -- especially in 9.1D where diagnostics are fairly limited.

 

[SSuhaib]

Thank you Thomas & Tom.

When I tried to add new extents and 'prostrct add' the extents were added but When I tried startup ( which does a crash recovery) the DB failed to start. Iam I missing anything here ?

BWT we did restored and applied the ai files which again resulted a few transactions lost.

Now the DB is started with -bithold & -bistall and as Tom suggetsed we are continously monitoring the active transactions to catch hold of mischevious program that is cause of this nightmare.

Thanks.

 

[TomBascom]

I have no idea why adding extents is failing. I suggest that you open a new thread regarding that and that you post more details. Specifically the error messages received.

 

[SSuhaib]

Thanks Tom.

The extents were getting added and we can see that by 'prostrct list' but when we proserve or try to truncate the DB fails. The corrupted database is deleted now. The messages were "memory voilation(49)" , "initating abnormal shutdown" and "the server terminated due to signal 3. (80)".

Thanks.