Progress 8.3E database / Mfgprod 8.5E ( heavily customized )
Running on Tru64 5.1B
The main database is about 60 GB
230 concurrent users
single AI extent ( a bit more than 1GB of AI data per day )
We actually do offline saves.
We are in the process of doing online saves using a combination of proquiet and filesystem snapshots. ( We will implement multiple AI extents )
Our goal is to shutdown the database 1 time a week during the week-end.
My question is about the BI extent behavior.
We actually have one BI extent with 9 x 64 MB clusters.
Someone said to me that when new clusters are created, progress (8.3E) is unable to re-use the previous clusters even if there's no more active transactions in them...
But I haven't seen any mention of that anywhere else. The documentation is not really clear on that point.
Where's the truth ?
Message was edited by:
marc michel
In general, clusters are always re-used UNLESS there is an active transaction that is using an early cluster in the ring. For a good details explanation about logging, read this engine crew monograph:
http://www.peg.com/techpapers/monographs/logging/logging.html
Thanks for the link.
Parts 8 and 9 of the document are what I needed.
If I understand correctly, when our database will be opened during 5 days, only potential very long running transactions could be a problem by preventing clusters to be freed.
Our BI file is monitored by Patrol, so we will have an alert before the fatal crash anyway ( 2GB limit ).
Thanks again for this quick reply.
Message was edited by:
marc michel