My company is building a private cloud and we are moving to cloudstack. As
we begun investigating the cloudmonkey CLI we found that the output was
slightly hard to read. I have begun working on some optimizations that I
think will benefit the community and I reached out to Rohit, who
recommended that I join this list and present my ideas. Here is what I am
proposing:
1. Add json output to cloudmonkey
I have accomplished this by adding a config parameter called display, which
can be set to json, tabularize, or default. I have removed the tabularize
parameter.
Justins-MacBook-Pro:cloudmonkey grudzien$ cloudmonkey list users
account=grudzien
{
"count": 1,
"user": [
{
"account": "grudzien",
"accountid": "b799783d-e5bb-460a-be0e-3966bd69edda",
"accounttype": 1,
"apikey": "*nokey*",
"created": "2013-03-27T16:09:17-0500",
"domain": "ROOT",
"domainid": "7e61c32f-9873-4944-947a-dcc00d3bebdc",
"email": "grudz...@gmail.com",
"firstname": "Justin",
"id": "265930bc-62ef-41f8-849c-e58593ca4b1f",
"lastname": "Grudzien",
"secretkey": "*nokey*",
"state": "enabled",
"username": "grudzien"
}
]
}
2. Add filtering as a standard parameter for all output types.
The only thing that has filtering now is the tabular output and grep breaks
the json.
Justins-MacBook-Pro:cloudmonkey grudzien$ cloudmonkey list users
account=grudzien filter=account,email,username,state
{
"count": 1,
"user": [
{
"account": "grudzien",
"email": "grudz...@gmail.com",
"state": "enabled",
"username": "grudzien"
}
]
}
3. Add color to the json output
I was thinking of colorizing the keys in the key/value pairs to increase
readability.
4. Move the color option from the config file to the command line.
There are two reasons for this. First, I want to be able to wrap a script
around cloudmonkey and not have to worry about colorization that will
impede me processing the output and second I think it would be more useful
to use the highlighting on demand rather than having to back out of the
shell to edit a config file.
5. Standardize messaging for the output types.
Right now certain kinds of messaging is presented differently for an output
type. For example, if I issue an api command that doesn't exist it displays
a generic error message, regardless of the output type selected. Ideally,
all output would be in the specified format.
I have the first two working and am planning on implementing the others as
I flesh them out. I will submit a patch when I feel it is ready. Any early
feedback on whether these changes will be useful to others is appreciated.
Justin
