Fog PXE boot multiple sites issue

[Wayne Workman]

I’m going to quote one of @Tom-Elliott 's old posts:

Why not use the Location Plugin to do the transfers for you?
Heck, if you update to the Development versions, you don’t even have to setup rsync tunnels.
You can install the Nodes how you see fit. I’d recommend, for your case, to install all the servers as “Full Servers”, and then once the installation is complete edit the /opt/fog/.fogsettings file to use: snmysqluser=‘fogstorage’ snmysqlpass=‘fogstoragepasswordfromfogsettings’ snmysqlhost=‘IP.OF.Main.Server’
This way, all the fog servers at all of the buildings communicate to a single server.
Then you create your storage nodes based on the information of the other fog servers.
Create the appropriate groups as necessary.
Assign the images to the groups you want the images to “cross” between.
That way you have a centrally managed server, with pxe boot setup locally at each building.
The location plugin will attach to the hosts that belong at that particular building.

Please give us details on your experience or thoughts, and please feel free to ask questions. We are here to help.

[abstert]

@Wayne-Workman This doesn’t seem to work as flawlessly as described by Tom.

We had to add manually go into the MySQL database on the master fog server and allow access from the storage nodes as adding the storage node username and password into the FOG website configuration menu didn’t do anything. We were then able to get remote pcs to boot into the the local storage node’s PXE menu, but when we click on the “Perform Full Registration & Inventory” the remote pc just reboots and launches whatever pre-install OS there is instead of running the registration process.

[Tom Elliott]

So there’s no trust in what I’m saying? I didn’t describe a full on configuration because I don’t feel it’s a requirement.

I may not know everything, but what I say is typically in a “technically” possible situation.

Essentially, instead of installing Storage nodes, install all the nodes you want tftp from as normal servers. After the install is complete, install the nodes to the FOG GUI as you would any other node. Ensure the other servers point at the same database. You can then manage the FOG Server from any one of the servers you installed, and they are all “Central” servers. Of course if you feel you have a better mechanism, by all means post it.

[Wayne Workman]

@Tom-Elliott This does make sense, it’s likely how I would want to set up a centralized site. Get all the bells and whistles and simple straightforward updates - and centralization too.

[abstert]

@Tom-Elliott
I didn’t mean any offense by what I said.

[Tom Elliott]

@abstert I know, remember i’m reading multiple posts a day trying to keep track of bugs and potential issues. When I suggest things, I’m not necessarily describing a fully set method to how things have to be done, just that they “should” work. I can’t test all possibilies though.

[Obsidian]

Guys,

Thanks for all of your help! I have tried what you suggested (reinstalled from scratch as a stand alone server instead of as a storage node), installed the location plugin on each one and set each one up as a storage node on the other, and the images are not replicating and I still cant seem to boot from PXE. Here is a copy of the .fogsettings file. Also I have verified my pxe boot menu options are setup as follows.

FYI as a storage node they do show they can see each other and so far there are no error messages in the image replicator logs.

Option 066 10.6.1.16
Option 067 undionly.kpxe

GNU nano 2.2.6 File: .fogsettings

## Created by the FOG Installer
## Version: 4580
## Install time: Tue 08 Sep 2015 05:32:45 PM CDT

ipaddress="10.6.1.16";
interface="eth0";
submask="255.255.255.0";
routeraddress="         option routers      10.6.1.1;";
plainrouter="10.6.1.1";
dnsaddress="    option domain-name-servers      10.6.1.11; ";
dnsbootimage="10.6.1.11";
password='e62765';
osid="2";
osname="Debian";
dodhcp="n";
bldhcp="0";
installtype="N";
snmysqluser="fogstorage"
snmysqlpass='fs43697453108';
snmysqlhost="10.5.1.16";
installlang="0";
donate="0";
storageLocation="/images";
mysql_conntype="MYSQLI_ASYNC";
fogupdateloaded="1";
storageftpuser="";
storageftppass='';
docroot="/var/www/html/";
webroot="fog/";
caCreated="yes";
startrange="10.6.1.10";
endrange="10.6.1.254";
bootfilename="undionly.kpxe";

Mod edited to use code box.

[abstert]

@Obsidian

What is the output of your replicator logs? Can you paste them here? You can go to your FOG site and go to the FOG Configuration Icon then the Log Viewer link on the left hand side. Be sure to select “Replicator” from the file menu.

[Wayne Workman]

@Obsidian Your FTP user and pass are blank… I’ve never seen blank credentials before, I am not sure it’s even possible to not use credentials for FTP in FOG…

storageftpuser="";
storageftppass='';

fyi, FTP is used for image replication. Also - please give us a screen shot of this area: Storage Management -> All Storage Nodes

And, while replicator logs would normally be very helpful in troubleshooting as @abstert requested, not having any FTP credentials is most likely the first thing that needs resolved. Then, yes by all means, give us Replication logs.

[Obsidian]

Okay so I found out that my PXE boot was an issue with the pc I was using for test I was able to get anouther pc to get the pxe boot menu fine.

Here is a screen shot of the config file of my main server that has my images on it. As far as the FTP user and pw they are blank on it as well. I think the system uses the fog username and password or the admin/root username and password for that pc for credentials.

I looked at the logs again to see what they looked like this morning and here is what I found. I found that there was an error logging in via ftp on the site 2 location. So I updated that the the root username and password and it is now transferring images.

Main Server:

Main Site Log files:

I am hoping that this means everything is resolved. Though I am not sure why site 2 does not have any thing in the log files. Also does this mean that I can upload an image to site 2 and it will be replicated to sites 1,3,4 & so on? Or should I only upload an image to Site 1?

[Wayne Workman]

@Obsidian What is set in all your storage node’s username and password fields? Anything?

[Obsidian]

@Wayne-Workman It has the user name and password of the Root user account for linux

[Wayne Workman]

@Obsidian said:

@Wayne-Workman It has the user name and password of the Root user account for linux

So you know, the username and password set for each of your storage nodes - these are the FTP credentials to be used with that storage node.

the installfog.sh installer creates a “fog” user account locally on the system automatically for you, and this account is password protected. This is also the account that is used for FTP and MySQL access. The installer - if it completes the installation - generates /opt/fog/.fogsettings and fills in those two fields I mentioned before with the username and the randomly generated password for that account. This way - when you upgrade in the future, the installer will setup shop using the correct settings.

For instance, my setup is using the local linux account called “fog” and a custom password I set on that account. I have this username and password set on my storage node and in my .fogsettings file.

There are many sets of credentials used with the entire package that we call “FOG”, and this is to allow you to modularize easily, to protect passwords to separate systems better, to separate low level access and root access, to lock down your system better. To give YOU - the administrator of your fog set up - absolute power over all pieces of FOG.

And, while using your root username and password will work obviously, I would not advise setting it up in this method.

[Wayne Workman]

@Obsidian One slip-up on your part, and your entire system is compromised… because all passwords match, and they all provide root access.

[Obsidian]

@Wayne-Workman That makes sense, I believe i have that written down since I redid this system and will update it with that. A couple of other questions.

Do you know with this type of setup can I upload an image at any site and it be replicated at all sites? and why the log viewer on site 2 is not showing anything?

[Wayne Workman]

@Obsidian said:

@Wayne-Workman That makes sense, I believe i have that written down since I redid this system and will update it with that. A couple of other questions.

Do you know with this type of setup can I upload an image at any site and it be replicated at all sites? and why the log viewer on site 2 is not showing anything?

The master node of a group should replicate to all other nodes in its group.
For uploading -
Say you create an image (or have an image already) that is set to Storage Group 1. Say that Storage Group 1 has a master node.

When that image uploads - no matter where it uploads from - the upload will be captured on the master node - and after that is complete - the image then replicates to other nodes in the group.

And for site 2’s log problems, it’s probably credentials related. is site 2 using the same version of FOG? what OS is site 2 running?

[Obsidian]

Wayne,

Site 1 is Ubuntu 14.04 Server & Fog Trunk 4557 & Site 2 is Ubuntu 14.04 Desktop on fog trunk 4580.

But now i have a new issue. The images have replicated and I can see them in the images folder on Site 2 however when I try to image a computer there are no images listed in fog. I have attatched screenshots of fog to show them not being listed and I was able to do an ftp to the system and list them in the directory.

[Tom Elliott]

Wait, what?

The second system is built as a server? FOG Doesn’t autopopulate the images stuff. that’s done by you.

[Obsidian]

Thanks Tom, I was thinking that it would for some reason, I went and added my image and now am imaging a system.

[Obsidian]

Not sure if I should post this with this topic or create a new one, if I need to create a new one just let me know. Hhowever I have been imaging the system (that I mentioned earlier in this post) and it has been imaging for over 3 hours now. The image is aprox 29GB and it has been transferring at aprox 130MB/Min it is plugged into the same switch as the imaging machine and by all calculations this should have been done in aprox 30 mins. Any thoughts or suggestions on why it is taking so long?