Re: Setting server and client for the same source code on same machine

[Denniston, Todd A CIV NAVSURFWARCENDIV Crane, Code JSXN]

Kiran Kamble wrote, On 06/17/2008 06:11 AM:
> Hi All,
>
>  
>
> Currently we are accessing our CVS server at our customer end. Since we have
> only one port open for committing to the customer server, multiple users
> cannot access thus one person has to manually take all the files from entire
> team and put them onto the client machine and then commit.
>
>  

I assume when you say "we have only one port open", you mean 'they have opened 
their firewall up so that only one of our IP addresses (i.e., one computer) 
can connect to their server'.

In that case the appropriate strategy would include one of the following:
Strategy A
A1) make the machine with that IP address a Kiosk machine at your company, 
i.e., all your developers can set down at it and log in.
A2 a) have all the sandboxes on a network drive that the Kiosk machine mounts.
or
A2 b) have all the sandboxes on removable drives that can be mounted on the 
Kiosk by the developers.
A3) have the developers each go to the Kiosk as needed to do checkouts, 
commits and updates.

Strategy B
B1) modify your firewall/Router to be state full & NATing and when ever one of 
your internal computers attempts to connect to the CVS server, the 
firewall/Router NATs the connecting machine to the IP address allowed to connect.

Drawbacks to A: developers have to make periodic trips to the Kiosk to get in 
sync with everyone else.
Drawbacks to B: figuring out if the firewall has timed out the last NAT so you 
can use it could be very confusing, if done automatically.  it might be more 
sane if you had a web page for users to go to and request a 2-5 minute window 
of having their computer NATed.

> Proposed Strategy:
>
> To overcome the issue we were thinking of making the client machine of
> customer CVS as the server for the local area network. This way all the team
> would commit to the "local server" from where it will be then committed to
> the customer server. This will make the process simple and automated
> reducing the manual errors in the process.
>
>  

CVSUP or something like it would do this better and be less error prone.

> In short can we use the same machine which will act as a client to the
> customer CVS server and a local server for the local area network for the
> same source code? Could there be any issues in this kind of a model? It is
> important from our perspective and we are keen on getting this implemented.

issues with your strategy: 1) you have to treat your customer's CVS as a third 
party source and 2) you still have to have a human pick out the changes and 
send them to the customer's CVS.

>  
>
> Regards,
>
> Kiran Kamble
>
>  
>
> Blue Star Infotech l*+91 9820 247630 l * +91 22 6695 6969 X881 l 6 +91 22
> 6697 3866l * address@hidden
>
>  
>
>   <http://www.bsil.com/> www.bsil.com - Where Partnerships Are Built on
> Trust
>
>  


--
Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane)
Harnessing the Power of Technology for the Warfighter
http://www.crane.navy.mil/custfeedback