Comment 26 for bug 1251784

Revision history for this message
Matt Riedemann (mriedem) wrote :

This is also showing up with about 50% failures:

http://logstash.openstack.org/#eyJmaWVsZHMiOltdLCJzZWFyY2giOiJtZXNzYWdlOlwiTG9jayB3YWl0IHRpbWVvdXQgZXhjZWVkZWQ7IHRyeSByZXN0YXJ0aW5nIHRyYW5zYWN0aW9uXCIgQU5EIHRhZ3M6c2NyZWVuLXEtc3ZjLnR4dCIsInRpbWVmcmFtZSI6IjYwNDgwMCIsImdyYXBobW9kZSI6ImNvdW50Iiwib2Zmc2V0IjowLCJ0aW1lIjp7InVzZXJfaW50ZXJ2YWwiOjB9LCJzdGFtcCI6MTM5ODM3MDY0ODU0N30=

Sounds like there are lock timeout issues with the database in neutron, maybe due to chatty agents, and I'm thinking neutron needs to wrap whatever is blowing up with a retry_on_deadlock decorator like nova does in it's db api layer.