Bug 1390199
Summary: | Launching instance sometimes results in having 2 IP's from the same network. | ||
---|---|---|---|
Product: | Red Hat OpenStack | Reporter: | Jeremy <jmelvin> |
Component: | openstack-nova | Assignee: | OSP DFG:Compute <osp-dfg-compute> |
Status: | CLOSED CURRENTRELEASE | QA Contact: | OSP DFG:Compute <osp-dfg-compute> |
Severity: | medium | Docs Contact: | |
Priority: | medium | ||
Version: | 9.0 (Mitaka) | CC: | amuller, chrisw, dasmith, eglynn, jlibosva, jmelvin, kchamart, mwitt, sbauza, sgordon, smooney, srevivo, vromanso |
Target Milestone: | --- | Keywords: | Triaged, Unconfirmed, ZStream |
Target Release: | --- | ||
Hardware: | Unspecified | ||
OS: | Unspecified | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | If docs needed, set a value | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2019-08-21 23:24:58 UTC | Type: | Bug |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Jeremy
2016-10-31 13:32:07 UTC
This bugzilla has been removed from the release and needs to be reviewed and Triaged for another Target Release. Hello, Seem to have another customer on OSP9 hitting this issue. Here is a better idea of how the problem happens: While booting 60 concurrent tenant instances we see something like the following: Some instances will fail on attempted build (overloading glance/swift?). The build will be retried according to the retry scheduler; however before the instances initial neutron port is released another is allocated. Both neutron ports remain associated with the rescheduled instance and the now successfully built instance has two IPs. This can can lead to a situation where you run out of IPs in the range during the multiple instance boot process if your range is close to the size of your concurrent boot. Instances that are still trying to build / reschedule then get an error because you have no more IPs remaining in your range. The upstream bug [1] was solved by this patch in Ocata [2] and backported to Newton [3], so I think we just need to backport [3] to OSP 9. [1] https://bugs.launchpad.net/nova/+bug/1609526 [2] https://review.openstack.org/#/c/393805 [3] https://review.openstack.org/#/c/396782 This thing where an instance gets rescheduled and gets two IPs has been reported before [1] (though that ended up getting closed EOL because it's OSP11). I'm not sure it's still happening in master, but definitely worth looking into. [1] https://bugzilla.redhat.com/show_bug.cgi?id=1469780 |