On 12/4/2017 1:51 PM, Binyamin Dissen wrote:
On Mon, 4 Dec 2017 08:33:44 -0600 Derrick Haugan <[email protected]> wrote:
:>Just an inquiry, to see if others have had issues with this products
use/abuse of XCF. BMC product is sending large bursts of XCF traffic between
systems, and the messages are the maximum size (62464). Over the past few months,
it has sent bursts like these (over 8K large messages in 1 minute interval), and
filled up XCF buffers on the sending system (transport class we use for large
message traffic). We are going to isolate the product's XCF group (DCPLEX$) into
its own transport class, so it cant block other users of the our general-purpose
large message transport class.
:>Product does reporting, and sends its results back to a focal point (sessions
used by DBAs etc) for analysis. Spoke with the vendor, they stated the options are
- a) turn off the XCF mechanism in the product or b) enable it for XCF
communication, but there is no control over what it may send (unlimited). These
seem like poor choice of options for tuning it. If XCF is disabled, DBAs would not
be able to assemble reports for DB2 using data sharing for multi-system operation.
I would ask if the XCF bursts are connected with the end-user/DBA requests for
data. Perhaps a screen that is refreshed frequently?
Bin,
I had the same problem with CA-ESP and GRS colliding. I created a
separate transport class for GRS and the problem was solved. Creating a
separate transport class is the only way to deal with this, currently.
Mark Brooks keeps talking about fixing transport classes, but nothing yet.
Regards,
Tom Conley
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
