Today I spent some time to investigate one of the problems which I found.
It was http://logs.openstack.org/47/653947/3/gate/tempest-slow-py3/629abee/testr_results.html.gz and as can be checked there, test failed due to paramiko.ssh_exception.AuthenticationException and that means that problem was no with connection to instance but with authentication to guest vm as cirros user.
So I looked into instance's console log and I found something like:
Sending discover... Sending select for 10.1.0.10... Lease of 10.1.0.10 obtained, lease time 86400 route: SIOCADDRT: File exists WARN: failed: route add -net "0.0.0.0/0" gw "10.1.0.1" checking http://169.254.169.254/2009-04-04/instance-id successful after 1/20 tries: up 19.66. iid=i-00000010 failed to get http://169.254.169.254/2009-04-04/meta-data/public-keys warning: no ec2 metadata for public-keys failed to get http://169.254.169.254/2009-04-04/user-data warning: no ec2 metadata for user-data
So it looks that reason of the problem was that public-keys wasn't provided by metadata service to the instance. What is also interesting is fact that connectivity to metadata in overall worked as instance-id was fetched from there properly.
So lets check now neutron-metadata-agent logs. I found there that request for public-keys was there: http://logs.openstack.org/47/653947/3/gate/tempest-slow-py3/629abee/controller/logs/screen-q-meta.txt.gz#_Apr_25_14_56_39_349166 And neutron metadata agent properly passed it to nova metadata server.
Lets then check nova metadata api logs: http://logs.openstack.org/47/653947/3/gate/tempest-slow-py3/629abee/controller/logs/screen-n-api-meta.txt.gz#_Apr_25_14_56_39_902289 Request comes there from neutron-metadata agent but I don't see any response for it. So maybe in this case it is some issue on nova's side? I think we should ask someone from nova to take a look into this.
Today I spent some time to investigate one of the problems which I found.
It was http:// logs.openstack. org/47/ 653947/ 3/gate/ tempest- slow-py3/ 629abee/ testr_results. html.gz and as can be checked there, test failed due to paramiko. ssh_exception. AuthenticationE xception and that means that problem was no with connection to instance but with authentication to guest vm as cirros user.
So I looked into instance's console log and I found something like:
Sending discover... 169.254. 169.254/ 2009-04- 04/instance- id 169.254. 169.254/ 2009-04- 04/meta- data/public- keys 169.254. 169.254/ 2009-04- 04/user- data
Sending select for 10.1.0.10...
Lease of 10.1.0.10 obtained, lease time 86400
route: SIOCADDRT: File exists
WARN: failed: route add -net "0.0.0.0/0" gw "10.1.0.1"
checking http://
successful after 1/20 tries: up 19.66. iid=i-00000010
failed to get http://
warning: no ec2 metadata for public-keys
failed to get http://
warning: no ec2 metadata for user-data
So it looks that reason of the problem was that public-keys wasn't provided by metadata service to the instance.
What is also interesting is fact that connectivity to metadata in overall worked as instance-id was fetched from there properly.
So lets check now neutron- metadata- agent logs. I found there that request for public-keys was there: logs.openstack. org/47/ 653947/ 3/gate/ tempest- slow-py3/ 629abee/ controller/ logs/screen- q-meta. txt.gz# _Apr_25_ 14_56_39_ 349166
http://
And neutron metadata agent properly passed it to nova metadata server.
Lets then check nova metadata api logs: http:// logs.openstack. org/47/ 653947/ 3/gate/ tempest- slow-py3/ 629abee/ controller/ logs/screen- n-api-meta. txt.gz# _Apr_25_ 14_56_39_ 902289
Request comes there from neutron-metadata agent but I don't see any response for it.
So maybe in this case it is some issue on nova's side? I think we should ask someone from nova to take a look into this.