Issues in XmppConnection cleanup lead to multiple test failures
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Juniper Openstack |
Fix Released
|
Medium
|
Nischal Sheth |
Bug Description
Couple of issues in XmppConnection cleanup have been identified when
trying to stabilize bgp_stress_test and bgp_xmpp_
- LifetimeManager is not used correctly - actor is enqueued to the
manager directly from the ManagedDelete method instead of invoking the
LifetimeActor Delete method. Given the right sequence of events this
causes problems when the connection is deleted from both above (due to
test TearDown) and below (due to the mock agent bringing down the session)
- XmppConnections that are rejected during dequeue (because there's
already a connection from the same IP) are not tracked properly. This
causes problems such as deletion of a different incarnation of the
connection (i.e. same IP but not the same data structure).
Both the above result in assertion failures in the LifetimeActor
destructor.
description: | updated |
Changed in juniperopenstack: | |
status: | New → In Progress |
description: | updated |
Changed in juniperopenstack: | |
status: | Fix Committed → Fix Released |
information type: | Proprietary → Public |
Fixed by https:/ /github. com/Juniper/ contrail- controller/ commit/ 8ef6604fd532f78 6ccc80ab3eb6b37 5ac1017177