Overflow imaging from storage node that's not a master in a storage group

[JGallo]

Re: Storage nodes not deploying images

@george1421 In the post I referenced, you mentioned that the storage node concept doesn’t allow for overflow imaging. I have a question which involves that concept. Maybe I have my current FOG system incorrectly configured or something is not working properly.

Here is what I have. I have one FOG server with multiple storage nodes across school sites. I use the location plug in. Each school has a location defined and a master node associated with it. With each location, i also have a “slave” node that is replicated from the master node of each site. I define the default storage group of an image and let the replication of the master node of that location go to the slave node of that location.

My question is, when I go to image a lab by utilizing the group management in FOG, why would only the master node deploy the image? Wouldn’t the slave node take on any overflow requests? Say I had 30 machines but the master node is set to the default of 10 clients and the other 20 are completely ignoring the slave node that should at least provide an additional 10 open slots. We just imaged a lab and my observation was exactly that the slave node, even though it has the image files there, the clients completely ignored the fact that the slave storage node was available.

I have Ubuntu 16.04 with FOG 1.5.4.8 on the replication branch. Im also testing something out probably next week now that we are at that point of our projects and switching to the replication branch to see if uploading an image that has been already defined and needs updating will not go into a replication loop. That’s something else but I’m curious as to why the slave storage node in a storage group is ignored during the imaging task with that location plug in.

Thank you.

[Wayne Workman]

@JGallo Is the slave at each of your sites assigned to the correct location?

[JGallo]

@Wayne-Workman No. I imaging that by having the Master node set in the location plug in that the storage group would handle the overflow. Do I have to set the location to both the Master and the slave node for overflow imaging to occur? If I do, when I go to register clients, would one or two of the same locations show up when assigning a location to them?

[Wayne Workman]

@JGallo said in Overflow imaging from storage node that's not a master in a storage group:

Do I have to set the location to both the Master and the slave node for overflow imaging to occur?

Yes. It’s not overflow, it’s just the two servers working as a team to do the imaging.

[JGallo]

@Wayne-Workman Ok. Makes sense now. Could you guide me on how to set the slave node to the same location as the master node? I’m still confused on how to do this exactly because I’m looking in the location plug in settings and in the main menu of the location management it has create new location. I don’t see a way to add the slave node to the same location as the master node. Thank you once again for you help.

[p4cm4n]

@JGallo slave node, do you mean a full install of fog? or a storage node install?
i can’t probably help you but i’m curious as to how you have it.

[JGallo]

@p4cm4n Both Master node and slave node in a storage group are storage node installs. The master node is the “master” for that particular storage group. Replication is not an issue getting image to slave. I’m having issues getting clients to pull from the slave once master node slots become unavailable.

[Wayne Workman]

@JGallo I took a look at my home server and I see what you’re saying. Go to the existing location, and set the ‘node’ field to ‘please select an option’ aka blank. Leave the storage group as it is. See if that works.

[JGallo]

@Wayne-Workman Thank you. I will try tomorrow and observe when we deploy another image to a lab that we have scheduled. We have a three day weekend so my apologizes for late response.

[JGallo]

@Wayne-Workman Interesting bug I think may exist. So I went into the location plug in and set the ‘node’ field to exactly what you said ‘please select an option’ and hit update then going to the location management page I no longer see a storage group set for the location. The strange thing is that going into the location it does show a storage group defined. It’s just when I go to the location management page that it doesn’t show. So by setting the ‘node’ field to blank, it appears that no storage group is defined at least according to the main menu of the location plug in.

Here is what it looks like. Hopefully you understand what I’m trying to describe.

[Wayne Workman]

@moderators @testers @developers can anyone confirm that the location plugin should/does support a storage group serving a location? As opposed to an individual storage node serving a location.

[Tom Elliott]

@Wayne-Workman I know that it does. Simply assign ONLY a storage group, and any nodes within the storage group will be available to the host in question.

[Wayne Workman]

@Tom-Elliott That’s what I instructed @JGallo to do, but when he does that, it unsets the storage group. Look at his last post.

[JGallo]

@Wayne-Workman @Tom-Elliott I just imaged one computer with the empty storage group in the location plug in and it pulled the image from the slave node. I’m going to leave it like that for now and when we go to image the lab later today, I will report. Like I said before, it is strange that the storage group is not visually represented on the location management page but once you click on the location, you can see that the storage group is defined.

[Tom Elliott]

@Wayne-Workman It doesn’t unset the storage group, it just doesn’t show the storage group under the list all item. Likely just a typo, but at the moment not a high priority thing.

[JGallo]

@Tom-Elliott @Wayne-Workman Worked like a charm. The lab imaged with 20 going on at once with both nodes active and remaining clients were placed in line to wait until slots opened up. Thank you both for all your help.

[Sebastian Roth]

@JGallo said in Overflow imaging from storage node that's not a master in a storage group:

Interesting bug I think may exist…

As Tom said, just a display thing. I got around to figure this out today and pushed a fix to the dev-branch. I don’t have a proper setup to test location deployments and all. It does fix the display issue for sure (tested) but I don’t find the time to setup and test the whole lot.

[JGallo]

@Sebastian-Roth Cool. We have several projects in the mix right now but once it slows down, I plan on upgrading to 1.5.5 since i haven’t had a chance to do so. Would that fix happen to be in the working branch by any chance or should i stay on the dev branch?

[Sebastian Roth]

@JGallo No the last release 1.5.5 doas not have the fixes. Will be in the next release to some in the next weeks I suppose. Nut sure when though. You definitely have the changes when you are on the dev-branch.

[JGallo]

@Sebastian-Roth Fair enough. I will switch over to the dev-branch on nodes and fog server once we’re done with these projects. I should be able to do it prior to our winter break. Thank you again for your help on this.