Image status fails to become active in test_list_image_filters

[John Griffith]

Have seen the following at least a couple of times lately in gate failures:

http://logs.openstack.org/48/96548/1/check/check-tempest-dsvm-full/682586b/console.html#_2014-06-10_09_35_06_587

[Matt Riedemann (mriedem)]

Yeah I'm also seeing this, was asking about it in #openstack-glance this afternoon. Saw it here:

http://logs.openstack.org/44/99144/1/check/check-tempest-dsvm-full/36fb63b/console.html

And see a lot of warnings in screen-g-api.txt which led me to this:

http://logstash.openstack.org/#eyJzZWFyY2giOiJtZXNzYWdlOlwiQ29ubmVjdGlvbiBwb29sIGlzIGZ1bGwsIGRpc2NhcmRpbmcgY29ubmVjdGlvblxcOiAxMjcuMC4wLjFcIiBBTkQgdGFnczpcInNjcmVlbi1nLWFwaS50eHRcIiIsImZpZWxkcyI6W10sIm9mZnNldCI6MCwidGltZWZyYW1lIjoiMTcyODAwIiwiZ3JhcGhtb2RlIjoiY291bnQiLCJ0aW1lIjp7InVzZXJfaW50ZXJ2YWwiOjB9LCJzdGFtcCI6MTQwMjU5NjY0NjA5MX0=

I'm wondering if those connection pool limit warnings are contributing to the timeouts on the tempest side. When this times out, qemu-system is the top CPU usage, not sure what else is hitting the API at the time of failure though.

[Dina Belova (dbelova)]

The same appeared for the Ceilometer gate too: https://review.openstack.org/#/c/102466/