| 2016-04-12 11:11:09 |
Igor Zinovik |
description |
Detailed bug description:
During deployment run plugin may need to calculate some data (e.g. translate
DNS names into IP addresses, convert vCenter cluster names to internal MoRef IDs, etc)
and store this data somewhere.
There is also a problem of repeating this conversion on each node, e.g. if we deploy 3
controllers and we need to resolve some DNS name to IP plugin will repeat this task
3 times instead of just fetch result from storage. More optimal way will be to make
conversion one time, e.g. on primary-controller, save result if fuel db and then fetch
it when plugin runs on second controllers.
Data should be tagged with some kind of marker lets call it 'deploy_id', so we can
determine the fact that data was updated.
Steps to reproduce:
N/A
Expected results:
No need to do some complex actions to determine which deployment run
is happening.
Actual result:
Ability to fetch data in puppet manifests.
Less deployment time.
Reproducibility:
N/A
Workaround:
Dirty hacks like storing deployment run number in file on each node. |
Detailed bug description:
During deployment run plugin may need to calculate some data (e.g. translate
DNS names into IP addresses, convert vCenter cluster names to internal MoRef IDs, etc)
and store this data somewhere.
There is also a problem of repeating this conversion on each node, e.g. if we deploy 3
controllers and we need to resolve some DNS name to IP plugin will repeat this task
3 times instead of just fetch result from storage. More optimal way will be to make
conversion one time, e.g. on primary-controller, save result if fuel db and then fetch
it when plugin runs on second controllers.
Data should be tagged with some kind of marker lets call it 'deploy_id', so we can
determine the fact that data was updated.
Steps to reproduce:
N/A
Expected results:
No need to do some complex actions to determine which deployment run
is happening.
Actual result:
Ability to fetch data in puppet manifests.
Deployment run takes a bit longer time.
Reproducibility:
N/A
Workaround:
Dirty hacks like storing deployment run number in file on each node. |
|
