Sometimes openstack services return 502 error code

[Sergey Murashov]

ISO: {"build_id": "2014-08-25_00-01-17", "ostf_sha": "907f25f8fad39b177bf6a66fba9785afa7dd8008", "build_number": "476", "auth_required": true, "api": "1.0", "nailgun_sha": "44876ddae29823449e0cbc59428aafa466cbbbc2", "production": "docker", "fuelmain_sha": "5d44e4173f3b15b5991737269fbfae71275a63da", "astute_sha": "0c1b16721f7810b1c400f51937ecc6f058be8cff", "feature_groups": ["mirantis"], "release": "5.1", "fuellib_sha": "31debf95128afa44a9991491eaff169625cb49b1"}

Steps to reproduce:
1. Install OS(HA, ubuntu, 3contoller, 1 compute, neutron gre)
2. Run OSTF

Actual result:
Some tests can fail with 502 status code

The most likely issue in HAproxy

[Vladimir Kuklin (vkuklin)]

There are intermittent 502 errors in haproxy logs. Nevertheless there are no errors in keystone logs. It looks more like an intermittent connection failure which happened only for this particular request. So the question is if the next run of OSTF exits succesfully. If it is true, then it is just a connectivity problem, that will disappear after connectivity is restored.

[Vladimir Kuklin (vkuklin)]

also it may be related to https://bugs.launchpad.net/fuel/+bug/1361993

[Sergii Golovatiuk (sgolovatiuk)]

502 usually means that there is a problem between HAProxy and backend. We may either increase timeout and buffer for backend responses.

timeout server 5m
timeout client 5m
tune.bufsize 32768

[Nastya Urlapova (aurlapova)]

Sergii, if this issue affect ostf does it mean that issue isn't medium priority, because it signal for us that for production cluster will be many problems.

[Sergii Golovatiuk (sgolovatiuk)]

I set it as medium as bug doesn't indicate the direct problem with cluster
or deployment. It's rather the problem with the test itself or potential
issue under very high load which requires slight tuning of settings to give
more time rather than timeout backend.

[Matthew Mosesohn (raytrac3r)]

Moving to invalid because of no activity while incomplete for 7 weeks.