nclements's dashboard

nclements set one of nclements' replies as an official response to "kernel = "/{location_of_kernel}/{kernel}" WTF" in CohesiveFT

nclements set one of nclements' replies as an official response to "How to use .cft file?" in CohesiveFT

nclements replied on November 16, 2009 17:45 to the question "How to use .cft file?" in CohesiveFT:

Hello,

When you untar your file you should get a directory, not a file.

For example, if your file was called "vmimage_name.vmware.cft.tgz" then you should end up with a directory called "vmimage_name". Inside that directory you'll have a couple of files -- 2 vmdk files and a vmx file.

The vmx file will be your configuration file; the vmdk files are the disk images. For ESX to run this you'll need to import the vmdk files and then open up the vmx file.

Hope this helps.

Nicholas

nclements replied on October 28, 2009 14:05 to the question "kernel = "/{location_of_kernel}/{kernel}" WTF" in CohesiveFT:

For Xen images we don't insert and supply a default kernel. Instead we let you provide it from the dom0. Since the chance of us guessing the name and location of your kernel correctly is quite small, we just use the placeholder in the configuration file.

Generally, when you've set up a Xen host, you'll have installed some kernels for use by the VMs. If not then the default dom0 kernel can also be used for the domU.

Let's pretend that you're using Ubuntu as the host machine and you've a kernel called vmlinuz-2.6.26-2-xen-686 in the /boot directory. Then you'd modify the configuration file so it looks like the following:

  kernel = '/boot/vmlinuz-2.6.26-2-xen-686'

  memory = 768

  name = "xen-ubuntu-test1-1097501-1256624352"

  disk = ['file:/xen-ubuntu-test1-1097501-1256624352.img,sda1,w','file:/swap.img,sda2,w']

  root = "/dev/sda1 ro"

  vif = ['mac=00:1C:79:##:##:##']

(MAC address will vary depending on your configuration file.)
You might also want to put in a line like the following (if it applies):

  ramdisk = '/boot/initrd.img-2.6.26-2-xen-686'

Hope this helps.

Nicholas

nclements replied on October 25, 2009 16:20 to the question "Webmin build not working as expected" in CohesiveFT:

Hello,

I'm not sure which bundle you used to install webmin -- there are two (that I can see, and I think both are available to community users), but neither one is a "CFT-Approved" build.

I've looked at the newest one (built a couple of days ago). It looks as if the creator simply uploaded the Debian package and hoped that our system would install it automatically; it doesn't.

If you're the creator of the package send an e-mail to "support @ cohesiveft.com" (without the spaces) and we'll try to help you build the package properly. If not then you might like to try to build your own package/bundle, in which case we can walk you through the steps.

If you'd prefer to just try working with the machine you've built then here are some steps that might help:

$ cd /opt/webmin
$ sudo dpkg -i webmin_1.490_all.deb

I'm not sure exactly where this will place the contents of the package (or, indeed, if the contents are valid). Assuming all is well, though, you should have a functioning webmin.

Note: It's possible that you might be told that you're missing dependencies; simply note the name of the packages it's telling you about and run:
$ sudo apt-get install {packages}

Hope this helps.

Nicholas

nclements replied on October 21, 2009 23:37 to the question "Apache control panel under the Elastic server manager" in CohesiveFT:

The idea of the "Has Control Panel" function is for those apps that have some sort of control/administration tool -- Tomcat, ActiveMQ, WebLogic, etc. for example. What you can do there is set a relative URL (and port, if necessary), then ESM will have a link that allows you to easily open it up in a new tab/window.

If you want to see this in action then check the checkbox and add what you want (control panel port, control panel URL, etc.). Then save and select the System Overview menu item and then choose the service you've updated. (Note: You may have to click the Refresh link also.) Now you should have a new link above the Start option: Administration Console. If the URL you created is a valid one then it'll open up in a new tab or window.

Nicholas

nclements replied on October 21, 2009 18:41 to the question "Apache control panel under the Elastic server manager" in CohesiveFT:

Hello,

There's a couple of steps that you need to take here.

First you'll need to configure things for Apache. The default ESM configuration doesn't actually link to the Apache configuration files (we did this for security reasons, but it does mean an extra step). To enable it you select "configuration" under the Apache menu item and click "Edit". In the "Configuration Files" section type "/etc/apache2/*.conf /etc/apache2/sites-enabled/*" (you enter a list of configuration files separated by spaces, wildcards are allowed). Click "Save".

Now you should be able to edit the Apache configuration files in ESM. Once you've finished restart Apache (from the System Overview menu).

Since you've already opened the port on the firewall you should now be good to go. There is one caveat here: If you're using EC2 then you need to make sure you've also opened the port on your security group for the running instance. (We've not linked up ESM to EC2's security groups because that would require either storing your EC2 creds on the instance image or having you enter them in each time.)

Hopefully that helps.

Nicholas

nclements replied on October 05, 2009 00:48 to the problem "Elasticserver fails with Eucalyptus Public Cloud" in CohesiveFT:

Hello,

Could you send an e-mail to support @ elasticticserver.com with the server detail URL that was sent to you via e-mail. It should look something like http://elasticserver.com/server/{some_details_here}. Set the subject to something like "Help with Eucalyptus" and we'll see what we can do to get things working.

Thanks,

Nicholas

nclements replied on October 02, 2009 17:15 to the problem "useless error occurred msg" in CohesiveFT:

Clive,

We're aware that the error messages are far from helpful to the end user. On the other side of things, the real error message wouldn't have been much help to you either. We are working on trying to get some useful information to the user when their builds fail (it's actually quite rare and painful when it does happen).

You should be getting an e-mail from our support people explaining what happened in some detail in a few minutes. (I was actually watching the builds when your build failed a short while ago -- it was a fluke conflict that has already been fixed and so your build should now go through if you try again.)

Sorry about that.

Nicholas

nclements replied on August 27, 2009 12:33 to the problem "New user "edit profile" link broken" in CohesiveFT:

Richard,

Thanks for the bug report. We'll get this fixed immediately. We have some updates coming out for the rest of the site in a couple of days and this will be in there.

Thanks again.

Nicholas

nclements replied on August 07, 2009 03:44 to the question "Build Server Programmatically" in CohesiveFT:

It's a coincidence that you asked this question today...we were just discussing this exact subject about 5 hours ago, and trying to decide when (and how) it will be available for use by our users.

We've been planning this for a while but it's not yet complete.

I can't give any specific details at this time, but I assure you that it will be coming soon. Just keep watching our blog at http://blog.elasticserver.com/ for details.

Thanks,

Nicholas

nclements replied on August 03, 2009 21:36 to the question "Update to JDK 1.6 automatically" in CohesiveFT:

Hello Nitai,

I've completed updating some of the packages and things look good. One thing you might want to do is rebuild your private bundle with the Sun Java JDK rather than the JRE. If you don't then you'll end up with both, though only one is going to be operational -- the other will be a waste of disk space.

Hope this gets you to the next level.

Thanks,

Nicholas

nclements replied on August 03, 2009 15:33 to the question "Update to JDK 1.6 automatically" in CohesiveFT:

Hello Nitai.

There are 2 ways to accomplish what you want -- the quick and easy way and the right way. The easy way is for you to create a new package as part of your installation, and have that make the changes. The right way -- and I'm working on this right now, and have been for the last couple of hours -- is to fix the packages so they do the right thing.

I'll keep you updated on the progress for the 2nd option.

Nicholas

nclements replied on August 03, 2009 00:04 to the question "Update to JDK 1.6 automatically" in CohesiveFT:

I just took a look at the build log for (what I think is) one of your recent VMs and it appears that our build environment has been overly productive -- you have both 1.5 and 1.6 on the machine.

Version 1.5 can be found in the usual Debian/Ubuntu location (and is found directly in the default path). 1.6 is to be found in /opt/java. What you can do is try the following:

• Change JAVA_HOME to /opt/java


• Add a link to /opt/java/bin/java into the default path.

Hopefully this helps.

Nicholas

nclements replied on June 30, 2009 12:38 to the question "How can I change keyboard layout ?" in CohesiveFT:

Hello,

Sorry to say but at this time we don't have a way to configure the keyboard layout at build time. It is one of the many items on our roadmap but I'm not sure of the timeframe.

One possibility, however, is you could create and upload your own packages that set things up the way you want at creation time. (I've not tried this to solve this particular problem so your mileage may vary.)

Thanks,

Nicholas

nclements replied on June 02, 2009 20:07 to the question "system updates hang when run through elastic server manager web gui" in CohesiveFT:

Clive, thanks. We've confirmed the problem and I've filed some bugfix tickets to get it fixed. It seems to be hanging on the background process while doing the apt-get install on a package, which is a newish problem.

I'm not sure of the timeframe on this fix (or on the installation of an alternative kernel), but we'll try to keep you posted.

In the meantime please do as you said -- run apt-get upgrade (or aptitude upgrade) to fix things.

Thanks again.

Nicholas

nclements replied on June 02, 2009 13:52 to the question "system updates hang when run through elastic server manager web gui" in CohesiveFT:

Thanks, this will help a lot.

As a heads-up, this might require an update to the Elastic Server Manager. However I should be able to send you a walkthrough on how to fix it manually if you like.

Nicholas

nclements replied on June 01, 2009 20:26 to the question "system updates hang when run through elastic server manager web gui" in CohesiveFT:

Can you be more specific? What updates are you seeing hanging? What files/versions? Thanks.

nclements replied on June 01, 2009 20:22 to the question "kernel requires features not on your cpu" in CohesiveFT:

We'll investigate further and see what we can do.

I would argue that it is an Ubuntu issue since they provide the kernels. We selected a kernel from their images based on their defaults. When you installed natively on your machine the install disks were able to run a bit of logic that identified certain specifics about your chipset, and an appropriate kernel was chosen. We don't have that luxury as we know little about the host hardware on which our VMs are to be run. But as I said, we'll see what we can do.

nclements replied on June 01, 2009 12:10 to the question "kernel requires features not on your cpu" in CohesiveFT:

Clive,

This is not a problem with the VM per se, but with the Ubuntu kernel shipped with the VM. All of our VMware VMs should work on versions of VMware Workstation (5.5 and above), Fusion (1.x and 2.x) and Server (ESX 3.5 and up), and should work on all the supported hardware.

We use a snapshot of Ubuntu Hardy Heron (8.04) that was dated several months after the original release, but it apparently had the same problem on bare-metal machines. For instance, see here: https://answers.launchpad.net/ubuntu/....

So the answer is yes, our VMs are intended to work on PIII, M series and older P4 series chipsets, if the host system works on them. However if the host system passes on information to the VM that would have caused the default install of the OS to not work then the VMs may not work.

Could you, perhaps, try Ubuntu 8.10? That may have better results for you.

Thanks.

Nicholas