Bugfix/support openstack metadata #186
Conversation
added logic for supporting openstack
| @@ -210,16 +211,21 @@ def deep_update(a, b): | |||
|
|
|||
| def get_provider(): | |||
| try: | |||
| logging.info("Figuring out my environment (Google? AWS? Local?)") | |||
| logging.info("Figuring out my environment (Google? AWS? Openstack? Local?)") | |||
| r = requests.get('http://169.254.169.254', timeout=2) | |||
| if r.headers.get('Metadata-Flavor', '') == 'Google': | |||
There was a problem hiding this comment.
I am wondering, may be on Openstack there is something similar available?'
Unique header for example.
There was a problem hiding this comment.
from what I can tell there is not. I ran the following to check
root@patroni-patroni-0:/home/postgres# python3
Python 3.4.3 (default, Nov 17 2016, 01:08:31)
[GCC 4.8.4] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import requests
>>> r = requests.get('http://169.254.169.254', timeout=2)
>>> r.headers
{'Date': 'Thu, 07 Sep 2017 22:51:27 GMT', 'Content-Type': 'text/plain; charset=UTF-8', 'Connection': 'keep-alive', 'Content-Length': '98'}
I did find that there is an openstack specific path
http://169.254.169.254/openstack/latest/meta_data.json
perhaps this would be a more reliable method of determining that a host is running on openstack.
| r = requests.get('http://instance-data/latest/meta-data/ami-id') # accessible on AWS, will fail on Openstack | ||
| if r.ok: | ||
| return PROVIDER_AWS | ||
| except (requests.exceptions.ConnectTimeout, requests.exceptions.ConnectionError): |
There was a problem hiding this comment.
According to the #183 it will raise socket.gaierror: [Errno -2] Name or service not known (fail to resolve instance-data) on openshift and therefore will to fix the problem
There was a problem hiding this comment.
@CyberDem0n the problem was actually discovered in plain Kuberenetes running on OpenStack. However, If a user was running OpenShift on OpenStack, they would likely encounter the same error.
There was a problem hiding this comment.
I understand, but before your changes if was failing in a different place, because socket.gaierror: exception hasn't been handled properly.
Now it will fail on the line 220, because DNS can't resolve instance-data into ip address.
On the line 223 we are trying to catch only connection error exceptions.
If you do a change:
- except (requests.exceptions.ConnectTimeout, requests.exceptions.ConnectionError):
+ except Exception: It will start working, but in addition to that I think it would make sense to change line 224:
- r = requests.get('http://169.254.169.254/latest/meta-data/ami-id')
+ r = requests.get('http://169.254.169.254/openstack/latest/meta_data.json')There was a problem hiding this comment.
@CyberDem0n I have implemented the changes you suggested, I removed instance-data and replaced it with the metadata server ip address which should resolve #191
|
👍 |
1 similar comment
|
👍 |
refs #183 add initial support for running spilo in kubernetes on openstack