Huge I/O load


on our tamino servers (RHEL-5.7) we have a a huge i/o load approximately every 3 hours, so at this point the server load is growing nearly 8.00 for some minutes and so the users claims the app is slow or freezing for a while.

We saw at this peaks there´re database operations

-rw-rw-rw- 1 tamino tamino 3.6G Nov 30 15:09 AAE00001.1I0
-rw-rw-rw- 1 tamino tamino 6.0G Nov 30 15:09 AAE00001.1D0
-rw-rw-rw- 1 tamino tamino 22M Nov 30 15:09 AAE000010.1T0

but i cant figure out what´s going on there. Are this internal jobs in the database? Maybe the “temporary working space” wass written in the database or something like this? How can we reduce this peaks or optimize this?

Any hints are welcome

Best regards

When data in the Tamino database is updated, the modified information is not written immediately. This information is kept in the buffer pool to minimize I/O activity when a block is updated multiple times. When the amount of modified blocks in the buffer pool reaches a given limit the blocks are written by a buffer flush.

If the time needed for a buffer flush is too long, the server parameter “flush limit” should be set to a lower value.

Dear Heinz,

thx this hint fix our problem :slight_smile:

Best regards