Archived
This forum has been archived. Please start a new discussion on GitHub.
Delay of OneWay-Calls
Hi,
we have got a problem with our implementation of an event service:
when the event manager wants to notify all interested enities, one thread calls methods over ice with oneway-Proxies on the remote objects. But if the remote method is very slow in processing, the event manager gets slower and slower and adapt himself to the slowest remote host, as other hosts are informed in the same slow time.
As I understand the manual, the caller shouldn't be influenced by the processing time, as it hands over the call to the transport layer and should return immediately. Or is the transport layer not able to buffer numerous calls? In our test the number of 15 delayed calls was not superseeded.
Wolfgang
we have got a problem with our implementation of an event service:
when the event manager wants to notify all interested enities, one thread calls methods over ice with oneway-Proxies on the remote objects. But if the remote method is very slow in processing, the event manager gets slower and slower and adapt himself to the slowest remote host, as other hosts are informed in the same slow time.
As I understand the manual, the caller shouldn't be influenced by the processing time, as it hands over the call to the transport layer and should return immediately. Or is the transport layer not able to buffer numerous calls? In our test the number of 15 delayed calls was not superseeded.
Wolfgang
0
Comments
-
We wrote an extensive FAQ on why oneway calls can block in our connections newsletter. See here for an online version http://www.zeroc.com/faq/onewaysBlock.html0