Rally scenarios for heat failed by 'Connection aborted'
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mirantis OpenStack |
Fix Released
|
High
|
MOS Heat |
Bug Description
api: '1.0'
astute_sha: cff84956fbe8ac8
auth_required: true
build_id: 2014-11-06_21-29-20
build_number: '82'
feature_groups:
- mirantis
fuellib_sha: ef9f6407d0d5553
fuelmain_sha: 0d3f16254760999
nailgun_sha: 172c1acadabf59b
ostf_sha: 9c6fadca272427b
production: docker
release: '6.0'
Steps to reproduce
1. Deploy 100 nodes cluster with 3 controllers in HA and 97 computes with Cinder with ISCSI with neutron GRE
2. Run rally tests
config
{
"kw": {
"runner": {
"type": "constant",
"
"times": 97
},
"context": {
"users": {
"tenants": 1,
}
}
},
"name": "HeatStacks.
"pos": 0
}
All tests failed with error ('Connection aborted.', BadStatusLine(
results
+------
| Action | min | avg | max | failures | total | result |
+------
| ~total | 13.171 | 13.511 | 15.253 | 97 | 97 | :( |
+------
stderr-0000
<class 'requests.
('Connection aborted.', BadStatusLine(
Traceback (most recent call last):
File "/opt/stack/
method_
File "/opt/stack/
self.
File "/opt/stack/
f = func(self, *args, **kwargs)
File "/opt/stack/
stack = self.clients(
File "/opt/stack/
resp, body = self.client.
File "/opt/stack/
resp = self._http_
File "/opt/stack/
**kwargs)
File "/opt/stack/
return session.
File "/opt/stack/
resp = self.send(prep, **send_kwargs)
File "/opt/stack/
r = adapter.
File "/opt/stack/
raise ConnectionError
ConnectionError: ('Connection aborted.', BadStatusLine(
stderr-0001
<class 'requests.
('Connection aborted.', BadStatusLine(
Traceback (most recent call last):
File "/opt/stack/
method_
File "/opt/stack/
self.
File "/opt/stack/
f = func(self, *args, **kwargs)
File "/opt/stack/
stack_id = self.clients(
File "/opt/stack/
data=kwargs, headers=headers)
File "/opt/stack/
resp = self._http_
File "/opt/stack/
**kwargs)
File "/opt/stack/
return session.
File "/opt/stack/
resp = self.send(prep, **send_kwargs)
File "/opt/stack/
r = adapter.
File "/opt/stack/
raise ConnectionError
ConnectionError: ('Connection aborted.', BadStatusLine(
or run rally tests
{
"kw": {
"runner": {
"type": "constant",
"
"times": 97
},
"args": {
"
},
"context": {
"users": {
"tenants": 2,
}
}
},
"name": "HeatStacks.
"pos": 0
}
All tests failed with error "HTTPException: ERROR: <html><body><h1>504 Gateway Time-out</h1>"
results
+------
| Action | min | avg | max | failures | total | result |
+------
| ~total | 60.131 | 60.391 | 61.404 | 97 | 97 | :( |
+------
stderr-0000
<class 'heatclient.
ERROR: <html><body><h1>504 Gateway Time-out</h1>
The server didn't respond in time.
</body></html>
Traceback (most recent call last):
File "/opt/stack/
method_
File "/opt/stack/
stack = self._create_
File "/opt/stack/
f = func(self, *args, **kwargs)
File "/opt/stack/
stack = self.clients(
File "/opt/stack/
resp, body = self.client.
File "/opt/stack/
resp = self._http_
File "/opt/stack/
raise exc.from_
HTTPException: ERROR: <html><body><h1>504 Gateway Time-out</h1>
The server didn't respond in time.
</body></html>
stderr-0001
<class 'heatclient.
ERROR: <html><body><h1>504 Gateway Time-out</h1>
The server didn't respond in time.
</body></html>
Traceback (most recent call last):
File "/opt/stack/
method_
File "/opt/stack/
stack = self._create_
File "/opt/stack/
f = func(self, *args, **kwargs)
File "/opt/stack/
stack = self.clients(
File "/opt/stack/
resp, body = self.client.
File "/opt/stack/
resp = self._http_
File "/opt/stack/
raise exc.from_
HTTPException: ERROR: <html><body><h1>504 Gateway Time-out</h1>
The server didn't respond in time.
</body></html>
I can't attach snapshot because
du -sh /var/log/
22G /var/log/
tags: | added: scale |
tags: | added: heat |
Changed in mos: | |
importance: | Undecided → High |
milestone: | none → 6.0 |
assignee: | nobody → MOS Heat (mos-heat) |
status: | New → Incomplete |
status: | Incomplete → Confirmed |
Changed in mos: | |
status: | Confirmed → In Progress |
status: | In Progress → Fix Committed |
Changed in mos: | |
status: | Fix Committed → Fix Released |
I am more than sure that is the problem of our current heat-engine deployment - there is only one of them running even in HA deployment, and it might simply choke on load, as the actual job of creating and deleting stacks is done by this single heat-engine.
This behavior of heat-engine was a workaround for (then undiscovered) bug https:/ /bugs.launchpad .net/mos/ +bug/1387345 that prevented us from actually using Heat in multi-engine mode.
Patches with the fix that will allow us to really use active-active heat-engine deployment are /review. openstack. org/#/c/ 133455/ (fuel-web/master, merged) /review. openstack. org/#/c/ 133263/ (fuel-lib/master, on review)
- https:/
- https:/