Race to delete shared subnet in Tempest neutron full jobs
Bug #1357055 reported by
Matt Riedemann
This bug affects 11 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
Fix Released
|
High
|
Eugene Nikanorov | ||
tempest |
Fix Released
|
Undecided
|
Salvatore Orlando |
Bug Description
This seems to show up in several different tests, basically anything using neutron. I noticed it here:
That's on a stable/icehouse change, but logstash shows this on master mostly.
I see this in the neutron server logs:
This query shows 82 hits in 10 days:
message:"delete failed \(client error\)\: Unable to complete operation on subnet" AND message:"One or more ports have an IP allocation from this subnet" AND tags:"screen-
Logstash doesn't show this in the gate queue but it does show up in the uncategorized bugs list which is in the gate queue.
Changed in nova: | |
status: | New → Confirmed |
importance: | Undecided → Low |
Changed in neutron: | |
importance: | Undecided → Critical |
Changed in neutron: | |
milestone: | none → juno-rc1 |
Changed in neutron: | |
assignee: | nobody → Mohammad Banikazemi (mb-s) |
Changed in neutron: | |
milestone: | juno-rc1 → kilo-1 |
Changed in nova: | |
status: | Fix Committed → Fix Released |
Changed in neutron: | |
assignee: | nobody → Salvatore Orlando (salvatore-orlando) |
tags: | added: juno-rc-potential |
Changed in tempest: | |
assignee: | nobody → Salvatore Orlando (salvatore-orlando) |
tags: | removed: juno-rc-potential |
Changed in tempest: | |
status: | New → Confirmed |
Changed in neutron: | |
importance: | Undecided → Wishlist |
Changed in neutron: | |
milestone: | none → liberty-1 |
status: | Fix Committed → Fix Released |
Changed in neutron: | |
milestone: | liberty-1 → 7.0.0 |
To post a comment you must log in.
e-r query here: https:/ /review. openstack. org/#/c/ 114359/
If this is mostly just neutron full jobs, this might make sense, i.e. those run with tenant isolation so maybe there are multiple tests running at the same time allocating ports on the same subnet and then when one test completes and tries to tear down it wants to delete the subnet which is still got one or more ports on it, so we fail.
Seems like we could just do a simple check in the nova neutronv2 API before trying to delete the subnet to see if there are any ports still assigned, or just handle the 409 and move on.