2013-08-17 23:28:11 |
Dmitry Mescheryakov |
description |
Right now we wait in instances.py:_configure_instances() for .ssh/id_rsa to be generated with timeout 10 seconds. Sometimes that causes cluster creation failure, which is painful especially in case of integration tests.
That is not a proper place for the check - it should be located in _check_if_up() where other similar checks are performed. Moreover timeout should be removed to stabilize the CI. Later we can introduce finely tuned timeouts for this and other checks. |
Right now we wait in instances.py:_configure_instances() for .ssh/id_rsa to be generated with timeout 10 seconds. Sometimes that causes cluster creation failure, which is painful especially in case of integration tests.
That is not a proper place for the check - it should be located in _check_if_up() where other similar checks are performed. Moreover timeout should be removed to stabilize the CI. Later we can introduce finely tuned timeouts for this and other checks.
An example of tests failing due to the issue:
http://jenkins.savanna.mirantis.com/job/gate-savanna-integration/700/consoleText
The specific line is
Can't start cluster 'ci-700-22' (reason: Error getting user private key) |
|