Description of problem: This is a copy from bug 1628235 for the OCP 3.10. Version-Release number of selected component (if applicable): v3.10.68;Upstream:v0.1.19 How reproducible: alway Steps to Reproduce: 1. Write down the connection count between the ansible-service-broker and service-catalog. You can access the node with the ansible-service-broker running on. Get its pid, and write down the connection counts. Like below: #ps -elf | grep broker #nsenter -t ${BROKER_PID} --net netstat -n | grep ESTABLISHED | wc 2. Do some privision/deprovision operations. And, do some sync operations between the ansible-service-broker and service-catalog, such as: #svcat sync broker ansible-service-broker 3. Check the connection counts again. #nsenter -t ${BROKER_PID} --net netstat -n | grep ESTABLISHED | wc Actual results: The count is increased. Expected results: The count should not increase. Additional info: The fixed PR is: https://github.com/openshift/service-catalog/pull/28
Hi Jian. There's some confusion on how I tracked the fix for this issue. I'm going to mark this a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1641791 as this is the same issue. For reference, the "sockets are never closed" issue is tracked by the following BZs: 3.9: https://bugzilla.redhat.com/show_bug.cgi?id=1638726 3.10: https://bugzilla.redhat.com/show_bug.cgi?id=1641791 3.11: https://bugzilla.redhat.com/show_bug.cgi?id=1641796 Sorry for the confusion. Please let me know if this doesn't make sense or if you would prefer to track this in another manner. *** This bug has been marked as a duplicate of bug 1641791 ***