Hi Aaron,

I have been working on the VCL Sandbox image out of NCSU repository. I was
reluctant to use the trunk as changes are made frequently on it and it will
be difficult for me to to keep up. What version of VCL is there on the VCL
Sandbox image and how would I go about and update the current code? Also
from a quick look at the trunk code it seems that there has been major
changes to the post_load routine as well as there is a file
"currentimage.txt" which is supposedly located on the image, I did not find
any documentation regarding this on the confluence, thus I am no sure where
and what this file should be and do.

The IBM Cloud is providing one IP address which is a public one and I
believe it is handled by dhcp. By the time the provision-module exits the
load() routine that IP address is recovered. At which point VCL is using it
to log into the machine and make changes to it such as create a new user
and set the proper permission's. I am not sure about the NAT.

I do not think that the option of "manualdhcp" will work as before the
image start up there is no way of knowing the IP address.

Best Regards,
Milen Paskov
WSTI Intern
mppas...@us.ibm.com


|------------>
| From:      |
|------------>
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
  |aaron_pee...@ncsu.edu                                                        
                                                                     |
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| To:        |
|------------>
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
  |vcl-dev@incubator.apache.org                                                 
                                                                     |
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| Date:      |
|------------>
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
  |08/18/2010 07:29 AM                                                          
                                                                     |
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
|------------>
| Subject:   |
|------------>
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|
  |Re: Setting up Base Image                                                    
                                                                     |
  
>--------------------------------------------------------------------------------------------------------------------------------------------------|





Hi Milen,

Make sure your working with the latest code from the repository. The
getdynamicaddress routine is no longer used because it was fetching the
loopback address as you see here.

Also can you explain a little more about the ip address provided by the
IBM cloud? It sounds like the public address differs from the private
address that vcl is aware of. Is it using NAT?

Also not sure if this will be of any use, but there is an option called
"manualdhcp" in which vcld will only provide the ip address listed in the
vcl database, instead of fetching it from the guest OS. This of course
requires the address to be pre-defined for that quest and not dynamic.

Aaron

> Hey Aaron,
>
> I found that the post_load() routine is called from the new.pm. But as
the
> image has not set the up external_sshd etc the commands in the routine
> fail, however those commands as far as i understand are only used hide
rsa
> keys and  root password. These action can be performed from my module,
and
> thus making the call for post_load() unnecessary. So I let the request
> continue even if the post_load commands fail to see what will happen.
>
> What I saw was that procedure continued without a problem and started
> logging into the machine to adding user etc. However when the connect
> button was available on the web-interface the IP that it gave was
> 127.0.0.1
> so i when back to the log and found the following.
>
> 1.1.1.1 represents the public ip of the instance, i replaced here with
the
> dummy one for security reasons.
>
>    2010-08-17 16:27:03|31471|6:6|new|utils.pm:getdynamicaddress(1605)|
>    PrivateIP address for ibm1 collected 1.1.1.1
>    2010-08-17 16:27:03|31471|6:6|new|utils.pm:run_ssh_command(6180)|
>    executing SSH command on ibm1:
>    |31471|6:6|new| /usr/bin/ssh -i /etc/vcl/vclsandbox_rsa  -l root -p 22
>    -x ibm1 '/sbin/ifconfig |grep inet' 2>&1
>    2010-08-17 16:27:04|31471|6:6|new|utils.pm:run_ssh_command(6262)|
>    run_ssh_command output:
>    |31471|6:6|new| inet addr:1.1.1.1 Bcast:170.224.175.255
>    Mask:255.255.240.0
>    |31471|6:6|new| inet6 addr: fe80::dcad:beff:feef:1510/64 Scope:Link
>    |31471|6:6|new| inet addr:127.0.0.1 Mask:255.0.0.0
>    |31471|6:6|new| inet6 addr: ::1/128 Scope:Host
>    2010-08-17 16:27:04|31471|6:6|new|utils.pm:run_ssh_command(6276)|SSH
>    command executed on ibm1, returning (0, "inet addr:1.1.1.1 Bcas...")
>    2010-08-17 16:27:04|31471|6:6|new|utils.pm:getdynamicaddress(1652)|
>    dynamic IP address for ibm1 collected: 127.0.0.1
>    2010-08-17 16:27:04|31471|6:6|new|new.pm:reserve_computer(1033)|
>    retrieved dynamic DHCP IP address from 127.0.0.1: 127.0.0.1
>    2010-08-17
> 16:27:04|31471|6:6|new|DataStructure.pm:_automethod(697)|data
>    structure updated: $self->request_data->
>    {reservation}{6}{computer}{IPaddress}
>    |31471|6:6|new| computer_ip_address = 127.0.0.1
>
> Now because I was able to monitor the progress from the IBM Cloud as well
> I
> found out the IP that I needed and was able to log into the image with
the
> credentials provided by VCL. So the only problem that I can see is that
> VCL
> fetched the wrong IP. Any idea why this happened?
>
>
> Best Regards,
> Milen Paskov
> WSTI Intern
> mppas...@us.ibm.com
>
>
> |------------>
> | From:      |
> |------------>
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

>   |Aaron Peeler <aaron_pee...@ncsu.edu>
>
>  |
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

> |------------>
> | To:        |
> |------------>
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

>   |vcl-dev@incubator.apache.org
>
>  |
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

> |------------>
> | Date:      |
> |------------>
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

>   |08/17/2010 12:18 PM
>
>  |
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

> |------------>
> | Subject:   |
> |------------>
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

>   |Re: Setting up Base Image
>
>  |
>
>--------------------------------------------------------------------------------------------------------------------------------------------------|

>
>
>
>
>
>   In the upcoming release, rc.local will be overwritten on image capture
> by the OS/Linux.pm module.
>
> a post_load script would probably be a good way to go until we can make
> the dependency on two networks optional.
>
> Aaron
>
> On 8/17/10 10:55 AM, Milen P Paskov wrote:
>>
>> Hey Josh,
>>
>> Regarding the first question I was able to modify the image so I can
>> ssh in as root.
>>
>> Regarding the fooling I was think more about making minor changes to
>> the script /etc/rc.local (suggested in the guide to created Base
>> Image), located on the image, as well as implement my own version of
>> post_load(). What I do not know is what needs to be done after the
>> image has started up before handing it to the end user. If some of the
>> action required different approach then the one already defined in VCL
>> i can rewrite them for my module.
>>
>> Best Regards,
>> Milen Paskov
>> WSTI Intern
>> mppas...@us.ibm.com
>>
>> Inactive hide details for Josh Thompson ---08/17/2010 09:24:24
>> AM--------BEGIN PGP SIGNED MESSAGE----- Hash: SHA1Josh Thompson
>> ---08/17/2010 09:24:24 AM--------BEGIN PGP SIGNED MESSAGE----- Hash:
>> SHA1
>>
>>
>> From:
>> Josh Thompson <josh_thomp...@ncsu.edu>
>>
>> To:
>> vcl-dev@incubator.apache.org
>>
>> Date:
>> 08/17/2010 09:24 AM
>>
>> Subject:
>> Re: Setting up Base Image
>>
>> ------------------------------------------------------------------------
>>
>>
>>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On Monday August 16, 2010, Milen P Paskov wrote:
>> > Hello Community,
>> >
>> > I have made some progress with the provisioning module that will
>> > communicate with IBM Cloud, now I need to set up the image on the
>> cloud
>> > properly so that VCL can access them.
>> >
>> > I am been following this documentation:
>> >
>> https://cwiki.apache.org/confluence/display/VCL/Create+a+Linux+Base
+Image
>> >
>> > I have come across two problems.
>> > 1. Currently the images are set so that you can not ssh directly as
> root
>> > user, but it is possible to use "sudo bash" once logged on. Is there
>> a way
>> > to modify the image so I can directly ssh as root as required by vcl?
>>
>> man sshd_config - look for PermitRootLogin
>>
>> > 2. Once the IBM Cloud provisions the image it has only access to one
>> > network (I am given one IP, through which I can connect) a public
>> one. Is
>> > it possible to fool VCL to use the public network to log in and make
> the
>> > necessary  changes to the image before giving to the end users?
>>
>> One of the backend guys will have to answer this part, but keep in
>> mind that
>> when you "fool" part of a system, you are often going against underlying
>> assumptions upon which the system was built.  This can result in a
>> long trail
>> of fooling one part, then another part, then another one, etc.
>>
>> Josh
>>
>> > Any help would be greatly appreciated.
>> >
>> > Best Regards,
>> > Milen Paskov
>> > WSTI Intern
>> > mppas...@us.ibm.com
>> >
>> - --
>> - -------------------------------
>> Josh Thompson
>> Systems Programmer
>> Advanced Computing | VCL Developer
>> North Carolina State University
>>
>> josh_thomp...@ncsu.edu
>> 919-515-5323
>>
>> my GPG/PGP key can be found at pgp.mit.edu
>> -----BEGIN PGP SIGNATURE-----
>> Version: GnuPG v2.0.14 (GNU/Linux)
>>
>> iEYEARECAAYFAkxqjIYACgkQV/LQcNdtPQNoBwCfXhiNXkiYJotVWID1QcaMQ9SP
>> xFoAniAl3Fq5etPpoVjKYAyAQa7PHLsA
>> =clPU
>> -----END PGP SIGNATURE-----
>>
>>
>
>
> --
>
> Aaron Peeler
> Program Manager
> Virtual Computing Lab
> NC State University
> aaron_pee...@ncsu.edu
> 919-513-4571
>
>
>


Reply via email to