One normal node, multiple storage nodes, location plugin and image deployment
-
Hello fellow FOGgers
I’ve had a break from FOG for a few months but now the company I work for is moving away from using MDT and instead using FOG for all image deployment. So I’ve been tasked with getting the remote imaging part working now.
I have a aingle storage node at head office with 2 storage groups setup, one of which replicates to the remote storage nodes.
I have implemented the location plugin to handle image deployment to the remote sites. What I would like to achieve is using the head office normal node for all host registrations and image uploading but sync the image files to remote storage nodes and have image deployment controlled from head office but pulling the image file from the local storage node at each site.
This does not appear to be working however.
Initial TFTP on the remote client shows the IP Address of the main fog server looking for default.ipxe and boot.php then pulling down /bzImage and /init.xz files from the main server. The transfer of /init.xz file takes several minutes. I can see in PRTG that there is a connection between main fog server and the remote client.
Partclone then kicks in to restore the first partition (/dev/sda1) of around 300MB. But then the second partition piece (/dev/sda2) fails with error:
Partclone v0.270 http://partclone.org Starting to retore image (-) to device (/dev/sda2) device (/dev/sda2) is mounted at error exit -
This then continues on to black screen with the following error
############################################################# # # # An error has been detected! # # # ############################################################# Image failed to restore ############################################################# # # # Computer will reboot in 1 minute # # # #############################################################
On the fog server GUI the task shows commecning but then hangs at 97.10% on the first partition.
The same image can be deployed withn the head office fine so the image itself is ok. File size of the image on both the main server and storage node is the same - I’m going to check an md5 hash shortly to double check as potentially image is corrupt on node…
The other thing I wanted to check was the correct usage of the TFTP is node entry in the location plugin.
Way I have this configured at present is the remote node DHCP server points to the IP of the main fog server (not the storage node). And I have TFTP is Node entry unchecked. I have also tried checking this box but makes no odds.
So 2 main questions - has anyone seen the error above re the image and have a possible answer as to cause and a solution?
And secondly what is the correct setup with DHCP scope options and the TFTP is node entry in location plugin to achieve what I’m trying to achieve?
thanks for any advice one and all.
regards, Kiweegie,.
Mod edited to use code box.
-
Is the image type resizeable?
Also - if you wanted an image to be shared and replicated between two storage groups - go to the image you want this for, click on “Storage Group” and just add another storage group.
The image first gets replicated to the master nodes of the storage groups the image is associated with - then the replication happens normally, flowing from master to regular nodes within the groups. Did that make sense?
-
Also - since you work for a company that produces revenue - please see what can be done about donating to the FOG Project. See my signature for details.
-
@Wayne-Workman HI Wayne, thanks for the reply. All images are set as Single Disk - Resizeable. The sharing between storage groups is not the issue - I have this particular image shared between the 2 storage groups I have currently. I don’t need to have each remote site in it’s own storage group do I? I presume the image replication still hits them if they are members of a storage group on which the master stores the image.
My issue appears mainly in getting the image to deploy an image from the main fog server in head office to a remote site but have the image itself pulled from that clients local storage node.
I wanted to double check that i had understood the setup with location plugin correctly and that I should still have DHCP scope pointing to the main fog server address (066/067 entries in DHCP) and if I neededed to have the TFTP from Node checked on the Location plugin for that remote client.
Ref the donation it’s definitely on my to-do list. I do give what I can to open source projects I use personally but it’s sometimes difficult to get the “powers that be” to buy into this philosophy. I’ll tackle with my management in due course though and see what can be done. You guys are doing awesome work and time “ain’t cheap”.
regards Kiweegie.
-
Quick update.
I’ve updated to latest SVN today on both the main fogserver and one remote storage node on which I’m testing.
For the sake of example here’s how I’ve set things up here for now - names altered to protect the innocent.
Head office:
Auckland fogservermain1 (normal), fogservermain2 (storage)
Remote offices:
Wellington fogremote1 (storage) Invercargill fogremote2 (storage) Dunedin fogremote3 (storage)
I’ve set DHCP scope on all sites to point to fogservermain1 (066/067)
Location plugin has been added and setup locations for:
Auckland, Wellington, Invercargill and DunedinThere are 2 storage groups setup - default and Auckland
All the remote sites point to the default storage group for images, Auckland is used for head office images only so that images are not replicated to the remote sites.
I’ve checked the fog user guide re location plugins here: https://wiki.fogproject.org/wiki//index.php/FOGUserGuide#FOG_Plugins
But there is not a great deal of detail. The plugin configuration only has a** name, storage group, storage node **and **TFTP from Node **entries which for Wellington remote node is set as
Location = Wellington Storage group = default Storage node = Wellington TFTP From Node = unchecked
Am I missing something fairly obvious? I’m assuming that all the remote sites should still have PXE boot pointing to the main fog server, in this example Auckland. And that unless TFTP is configured on each remote site (I’ve not done this) then to leave TFTP From Node unchecked and the Storage Node entry itself is how the remote site knows which server to pull the image from.
Anyone who has successfully set up remote site imaging using the location plugin, could you chime in here to improve my understanding please?
Until todays update of the SVN version imaging remotely was dying a death. It is now imaging successfully however the image is being pulled from the Auckland main server not Wellington remote server still. I suspect I’m missing a very small step here but would appreciate a nudge in the right direction if possible please.
regards, Kiweegie
-
@Kiweegie You’re almost correct.
The TFTP Checkbox is not as you might expect.
I don’t have a way to ensure that tftp from storage node is actually going to include PXE/iPXE stuff. The TFTP From node is simply to ensure the bzImage/Inits are downloaded from the location as well.
-
Thanks Tom, looks to be the missing piece in my particular puzzle. I now have everything set up as before except with TFTP From Node checked and image deployment remotely works pulling image from the local storage node at each site.
Many thanks, Kiweegie.
-
Save to solve then?
-
@Tom-Elliott absolutely, was looking for how I could mark resolved this side funnily enough,