[ckan-discuss] Installation options

[Simon Whitehouse]

On 15/02/2012 10:36, Rufus Pollock wrote:
> On 15 February 2012 10:31, James Gardner<james.gardner at okfn.org>  wrote:
>> Hi Simon,
>>
>>
>>>> First post to the list, so hello, I'm Simon and I currently work in
>>>> Digital
>>>> Birmingham. I've been asked to look at open data platforms for the city
>>>> and
>>>> so have spent a bit of time looking at ckan recently after chatting to
>>>> Mark
>>>> Wainwright at UK Gov Camp last month.
>>>>
>>>> I've been following the package installation instructions here:
>>>>
>>>> http://docs.ckan.org/en/latest/install-from-package.html
>>>>
>>>> and firstly chose to install to Amazon EC2. That went really well, but
>>>> the
>>>> small version is going to cost me $1/day and I'd rather not pay that.
>>>> I've
>>>> installed ckan under a virtual machine, but that is quite painfully slow
>>>> on
>>>> the machine I have to use for it.
>>>>
>>>> I did see that there had been a discussion previously about creating an
>>>> Amazon ami file for the micro EC2 instance and I wondered if that had got
>>>> anywhere. There are a number of reasons why I'd rather use the Amazon
>>>> install, not least that I can deploy it for others to see a lot better
>>>> than
>>>> on a virtual machine running under Windows.
>>> Just to clarify, is the reason you can't use the micro is because we
>>> build for 32-bit? If so I would imagine (though could be wrong) that
>>> you could install the 32-bit on 64-bit architecture without anything
>>> going wrong (we may need to rebuild CKAN as no-arch but that shouldn't
>>> be too hard).
>>>
>>> @James: could you comment on this?
>>
>> We did talk about setting up an AMI, but as you found out, it is fairly easy
>> to install from packages anyway, so we never went that extra step.
>>
>> The packages are built for architecture "all" so work on both 32bit and
>> 64bit systems. EC2 micro instances really are quite small and as CKAN has
>> become more sophisticated it requires slightly more resources. I like to
>> give a production CKAN instance at least 1GB RAM to ensure it runs smoothly,
>> although you can use less, the performance won't be as good. If you want to
>> experiment with the micro instance, you would just follow the same approach
>> as for the small instance, but you'd need to find a suitable Ubuntu AMI (I
>> don't know of one myself) and choose micro instance during the setup. I'd
>> recommend a small instance backed by EBS though. Micro instances don't
>> support EBS as far as I know, so if Amazon rebooted a micro instance for any
>> reason, you'd lose all your data - probably not something you want.
> Really useful info James. Just to clarify, micro instances definitely
> do support EBS (in fact it can only be run off EBS ...).
>
> Rufus
>

Hi James and Rufus

Thank you for the helpful suggestions. I tried installing from a number 
of the Ubuntu ami files that Canonical make available, but each of them 
failed one way or another. So, I've decided to bite the bullet and load 
the ami file that is suggested in the Install From Package notes.

So, now I want to load some sample datasets and before I dive in to try 
and learn how to do this through the API or Harvester I wanted to upload 
a dataset through the CKAN front end. Now, I've taken a look at the 
notes here

https://github.com/timrdf/datafaqs/wiki/ckan

and they indicate that I ought to see an option to "Upload a File" in 
the Add a Resource section under >Edit>Resources on the dataset. 
Unfortunately, I am not seeing that option when I add a resource. For 
example

http://ec2-46-51-165-218.eu-west-1.compute.amazonaws.com/dataset/edit/test0#section-resources

I've tried giving author, editor and reader authorisation to the admin 
user, but that doesn't give me access to the "upload a file" button. 
I've done a couple of quick searches on the site but not turned up 
anything. If it's in the instructions then it's a part of it that I 
haven't got to reading yet.

Any pointers on where I'm going wrong here would be much appreciated.

All the best
Simon