Transcript

Introduction. Brief Review, Agenda and Introduction to Control Domains

Solaris Training Rapid Deployment of Logical Domains Part 1

 

[music]

 

>> Mick:  Welcome, everybody – good day to you all – to this one-hour webinar on Rapid Deployment of Logical Domains.

 

[pause]

 

There are quite a lot of practical elements to this particular webinar to demonstrate how rapid deployment works. But I’m going to start off by just covering a little bit of the basics again. We did run a webinar a week or two ago on how to establish the Logical Domains control software on a T-series server. But in case you missed it, I’m going to do a few minutes review just to remind everybody and to put things into perspective so you can understand how the machine is configured before we actually start the deployment of another Logical Domain.

 

[pause]

 

We could call this webinar “How to Create a Solaris Operating System Instance in Minutes Rather than Hours.” That’s not necessarily true if you have a number of different types of configuration. What if you’re rolling out the same type of operating system instance numerous times? Then what we’re covering in this webinar is very, very suitable.

 

Now, as you probably realized, the Logical Domains facility runs on Oracle’s T-series. There’s also known as Core Multi-Threaded (CMT) on what was inherited from the Sun product line and includes Ultrasparc processes going through T1, T2, the T3 series, and the T4 which begin shipping in December as far as we can make up from Oracle.

 

[pause]

 

These servers support processor chips which have numerous cores. Each core has a number of threads. Each of which is regarded as a separate CPU, which I can demonstrate actually. If I come into this window on the right-hand side – this is actually logged in to a T-series server. And if I do psrinfo which is a standard operating system command that you’re fairly familiar with, this will show the processes assigned to this particular system which in fact is zero to seven. As far as the operating system is concerned, each of the threads is a separate processor.

 

The seminar is aimed at people who have some Solaris administration experience and probably also expecting you to have a little bit of an understanding of the virtual technology that’s available on systems these days.

 

[pause]

 

We’re using a SPARC system T5120 which you can see here, hostname of whale. It has 8 cores of 8 threads on each core, so 64 apparent processes and a certain amount of memory out of this PCI slots and so forth. Using the Logical Domain software and the T-series architecture, we can create logical divisions of the hardware known as Logical Domains.

 

So we can effectively create physical systems as it were with a certain number of processes, disk space, and so forth. Each one, of course, runs as independent operating system. You can run your applications within it. You can patch each one separately. You can use it as a test and development database very easily and it truly is an independent hardware resource.

 

[pause]

 

Of course things like PCI buses and the network interfaces can also be assigned to various components of the domain, usually as virtual devices.

 

[pause]

 

What we’re going to look at then today is a quick review of control domain configuration, the control domain being the master domain, if you’d like, the initially installed Solaris operating system in which the Logical Domains’ manager software has been installed. Using the Logical Domains’ manager software, we can create other machines, the actual guest domains and so forth. The system provides us with the ldm command to allow us to actually do that.

 

Then we’ll look at creating a guest domain, getting it started, accessing the console of it which actually you can see over here. That’s one already done. I have to boot the guest domain.

 

What we’re going to do is build something called emulated volume which is the disk backend to the guest domain. We’re going sys unconfig the guest domain, but if you’ve had some Solaris experience, the sys unconfig command that you’ll know completely strips away the identity of the machine. So once you boot it the next time around, it stops and asks you all the questions about what’s the host name, what is the geographic region, what is the IP address, and all this stuff.

 

Once we’ve sys unconfig the system, we’re going to halt it. On the control domain, we’re going to use the ZFS facility to snapshot the entire operating system and clone it to create copies to allow extensive rapidly deploy new guest domains very, very quickly.

 

[pause]

 

I’m actually going to do this. Once we’ve got the domain configured, I’m going to show you how we actually go about creating an entire domain based on a clone. So you’ll see it all in practice as well.

 

There will be breaks when some of the practical will take a little while and that will then give you the opportunity to ask a few questions and for us to maybe fill in a few details of other facilities that are related.

 

[pause]

 

SkillBuilders is involved in this technology quite heavily and there’s just a note in the slides to say that SkillBuilders have all the background, the resources, the qualifications in Solaris and much experience with the T-series, so if anything is of interest that comes out of this webinar and you’re interested in the technology or you’d like to know more about it, feel free to call somebody on the number that is given here in the notes.

 

[pause]

 

Back to the control domain. When you’ve initially set up your T-series server – take it out of the box, if you like, rack it up and boot it, it normally comes configured with the Solaris operating system. But you can actually install your own if you prefer to do so. You can build it just like any other Solaris system. So from the start you can just build the thing, not bother with the Logical Domain software and have goodness knows how many CPUs and memory to use. But of course the main strength of these systems is the ability to create multiple machines within it and you you’ve effectively got a data center in a box with your database applications, your web services, all your development, your testing all in one physical space.

 

Just to let you know, you can build the initial operating system if you want. The normal way that you would build any SPARC system, you can use a DVD if the system has one or you can set up network booting. There’s an example of booting over the network typically to perform an installation.

 

[pause]

 

Now, once the control domain is up and running, you would then install the Logical Domains’ manager software. We did go through this on the last webinar, so I haven’t repeated the instructions. What I’ve shown here is once you’ve got the Logical Domains’ manager software installed, you then have to configure virtual services to support the services you’re going to provide the guest domains.

 

Here’s an example. Again, just a brief view of what I had to do to configure the control domain, creating a disk service – ldm add-vds primary-vds0 – the name of the default domain is called primary. A bit like the system is also the global zone and you can have other none global zones. When it comes to Logical Domains, the primary domain is like the global zone, if you like. You do everything from the primary domain to configure or change the other domains on your own.

 

[pause]

 

This is adding a virtual disk controller. You’ll see later how we can add disks to it.

 

Copyright SkillBuilders.com 2017

×
Transcript

Control Domains Demonstration, Guest Domains, IO Domains and more…

Solaris Training Rapid Deployment of Logical Domains Part 2

 

>> Mick:  This is setting up a virtual terminal server through which I gain access to the Logical Domains once they’re created using Telnet. I defined the range of ports that I use to access the system.

 

Here’s an example. Telnet localhost 5000. Here I’m on my control domain, the same machine I’m on here awhile. I want to talk to my first Logical Domain which is given the Telnet port of 5000. I say, telnet localhost and then the port number.

 

If I come back to this window, you’re thinking, “How would I know what the port is?” I can do things like an ldm list and here I’ve got a domain called hpf1 and the console is port 5000, so I can see quite clearly what port to access. And having accessed it before, I’ve got into the OK prompt of that virtual machine. I can run all the normal commands that you would expect at that particular juncture.

 

[pause]

 

Hit the banner. I can actually assign the MAC address.

 

[pause]

 

But the LDom software assigns a MAC address for me. I can do things like show disk and have a look at the actual disk drives. One of them would be the operating system disk and the other disk there, A, is a DVD from which I’ll show you how to assign a DVD of the Solaris image and also be able build the machine without using the network.

 

All the usual commands are available.

 

[pause]

 

Here I’m adding a virtual switch. The Logical Domains software has some very powerful facilities. I’m creating a virtual switch through which the Logical Domains can communicate within the physical machine and which itself can also communicate with the outside world.

 

Here I was trying the actual physical port of the system that I want to be used for that virtual switch to talk to the outside world. As I’m configuring all these things I can list the services as we go along, ldm list-services, to keep a track of what’s going on. Obviously, the list gets bigger the more work that I do.

 

[pause]

 

As well as configuring services, I also need to configure the actual physical resources that the control domain is going to use. Rather than using the MAU which is the default when you build it, I can actually trim down the resources the control domain is going to use.

 

Again, I did cover this in the last webinar. But in case you’ve not seen that, here I’m setting a mathematical unit assigning one unit. In this particular machine there’s a mathematical unit per core of each of eight threads. I’m setting one mathematical unit and then eight virtual CPUs, which is equivalent to one core. In fact, I can assign cores as well.

 

[pause]

 

Here I’m assigning memory but I have to do this in a special way with the control domain because the LDom software may complain about the fact that I’m changing the configuration of the control domain. So I have to delay the reconfiguration until I reboot the control domain. That’s why this command is run, ldm start-reconf primary.

 

It says, “Any changes I’m now going to make, please delay them until I reboot the machine,” and then everything will work nicely instead of getting error messages. I’m assigning two gigs of memory. The control domain is only going to be a control domain. I’m not going to run an application, therefore I don’t need a lot of resource. If any case, I can dynamically change the amount of memory as I’m running, which is another great feature of Logical Domains.

 

I would reboot my domain and then I can see if I come back to the domain, I can do ldm list-bindings, the name of the domain, and I can check to see that the resources are as I expected. I can see a little summary at the top, two gigs of RAM, eight virtual CPUs. If I look through the listing, that accompanies that. You can see there’s the mathematical unit, things like what the boot device is, etc., and the I/O devices.

 

It’s possible to split depending on the nature of the hardware. I could split PCI buses across different domains. But that’s for another webinar a little bit later in the future.

 

[pause]

 

So the control domain is now set up with these resources. I have to enable the virtual network terminal service daemon to allow me access to the port like I did here with Telnet support 5000.

 

Then I can start thinking about guest domains. I need to know before I start creating guest domains because I haven’t created them here.  I’d obviously know pretty well what resources I had, but I could do ldm list devices which would show me basically what I’ve got left, what’s free at any given time.

 

[pause]

 

I’ve got a few cores left, a few mathematical units. If you’re wondering where the others went, I’ve actually already created a domain with hpf1 which we’ll be looking at shortly.

 

We’re going now look at creating the domains, assigning resources, binding it which is basically then assigning all the resources from the control domain and getting the domain started and installing the Solaris operating system within it.

 

Just to let you know a little bit more about the capability of the T-series server, there’s a lot of extra facilities that I haven’t mentioned especially with the more recent T-series servers like the T3s and T4s. It’s quite common to create domains through which you can split the input/output from the machine, and then you can create failover of devices like a PCI bus that maybe has an HBA card configured within it, so you can then create multipathing of disk to your SAN or whatever devices you’re using.

 

If one of the domains goes down through some operating system fault, which obviously is unlikely but covering all angles, the other domain will automatically then be used for the I/O path of the disk. And which is what we’ve done on this T3 server that I’ve been talking about. Some fairly complex configurations are possible.

 

This particular server runs a control domain and other domain is used just to handle failover of I/O and balancing. It has six guest domains, each of which is running a fairly substantial Oracle database. I think the system has got 256 processes all together and a vast amount of memory.

 

The nice thing is if you consider the tuning of these things a little bit down the line and you find that one of the systems is using a bit more process or power than some of the others, you can dynamically allocate processes to that system and remove them from something else. You can even now, with the latest version of the Logical Domain software, do the same thing with the system memory.

 

[pause]

 

In this particular case, with this T3 that SkillBuilders are configuring and looking after, each guest domain also as well as having disk multipathing uses Solaris IPMP through two different domains so that each one has a couple of network interfaces, one of each of which is configured in the separate I/O domains. Again, if there’s an I/O domain failure or control domain failure, not only do the disk paths failover, that’s also true in the networking.

 

In fact, you could split the PCI bus between machines and you can split individual PCI slots between machines, again depending a little bit on the hardware. That’s not always guaranteed but the T3 too certainly has the capability.

 

Copyright SkillBuilders.com 2017

×
Transcript

Guest Domains Demonstration

Solaris Training Rapid Deployment of Logical Domains Part 3

 

>> Mick:  As I mentioned, before we go on and create guest domains, we can do a list-devices and see what we got left.

 

[pause]

 

We also have to consider when we build a guest domain where we want to put the operating system. It could be on a separate physical disk. With ZFS, we could create what’s called an emulated volume which will look like a device. So grab a certain amount of space that create a device entry for it and then we can use that as the actual disk backend, which we’re going to do to rapidly deploy our domains a little bit later.

 

If we have UFS file system, we could use a mkfile command and create a fixed file and use that as the operating system backend. You could use SAN volume manager metadevices, although that’s probably a little bit unlikely. With ZFS technology, it sort of overtaken volume manager a little bit.

 

ZFS volume, emulated volumes are probably the best way to go. You can use Veritas. We could use any attached SAN, LUNs, iSCSI devices. If you’re lucky, you might have one of the new Oracle 7000 Series Unified Storage Solutions, which will provide massive amounts of storage for virtual machines at amazingly fast speeds with any product, all you can think of. Very, very easy to manage.

 

Those two things, something like T-series have and Unified Storage Solution would be an ideal match. In our case, we’re going to use ZFS emulated volume. If we look at a zpool command, we can see what the disk capacity we’ve got.

 

[pause]

 

On the control domain, I’ve got a mirrored arrangement. I’ve got 123 gigabytes for that.

 

[pause]

 

I’m doing a ZFS list so I can see the actual ZFS file system available.

 

[pause]

 

What we’re going to go about things is to create an emulated volume which basically grabs some of the space and makes it work as though it’s a physical disk, if you like, and then we can assign that emulated volume to the logical domain when we set it up. Then we can build the Logical Domain that Solaris thinks that is a real disk and we can build Solaris on it.

 

[pause]

 

One nice thing we can do when we’re creating an emulated volume is use a sparse option. We create a volume, let’s say a 500 gigabytes, even a 119 available gigs and it doesn’t actually use any space. It only ever uses the amount of space that is written to it. If in future months or years, the thing looks like it’s going to have flow the actual physical availability, using the zpool facility, we can add more disk devices to the flow without having to worry about all the underlying things that we’ve created like the emulated volumes and they will just naturally carry on growing as we patch, as we upgrade and so forth. So it’s a great way to go about things.

 

[pause]

 

Once we’ve created this operating system, we will probably patch it to make sure that it’s nicely up to date.

 

[pause]

 

The steps we would do, we’d create the guest domain on the aforementioned emulated volume, install the Solaris operating system within the new guest domain and then log into it and do a sys-unconfig.

 

Now here we’ve got the domain that’s already built, so I’ll just boot it up. While this is going on this might be a good opportunity to pass over a couple of questions.

 

>> Dave:  Hi, Mick. Yeah. There’s a question in the queue and that is, what command is used to create a ZFS volume?

 

[pause]

 

>> Mick:  That’s actually zfs create with -V option and very shortly, I’m going to actually show you how those done.

 

>> Dave:  Great. Thanks. Another question that came in to the queue I can answer and that is, will the webcast be made available after this session?

 

Yes, it will. We’ll send you all a link to the recorded version of this webcast so that you can review it as often as you’d like and share it with your colleagues.

 

>> Mick:  Thanks, Dave. If you all now look at the terminal window, you can see that I’ve got a login prompt. So my first Logical Domain that I’ve created and patched with the hostname seal is now booted nice and cleanly by the look of it. And when I log in it’s a standard Solaris build, I’ve patched it it’s conforms to a nice clean pattern.

 

Now what I’m going to do – just to show you it has a network identity, it has a time zone set, it has a hostname (sealed in this case). I’m going to take all that away now and do a sys-unconfig.

 

[pause]

 

Don’t do this at home, as they say. This will completely destroy all the identity of the machine. If you’re wondering exactly what it does, there’s a very good landing page if you do sys-unconfig. It lists exactly all the changes that it’s going to make. It will halt the machine.

 

>> Dave:  I’d like to take this moment to remind everybody to keep their chat windows open. I’m putting some questions out there for the group. I’d like to know if anybody’s running Oracle database on Solaris. If so, what version of Solaris are you running? You can chat that back to us, that’d be great. Thanks.

 

[pause]

 

>> Mick:  Okay. Here I’m now shutting down the initial Logical Domain. So there’s nothing special about the way I built it. It’s going to make a nice template to provision all the other domains that I’m going to create.

 

[pause]

 

The emulated volume that I created you can actually see here in this window. I’ve created a zfs file system or dataset, if you’d like, with the LDom files just to keep things nicely separated. Within that I did – this is in the notes very shortly – zfs create-V.

 

[pause]

 

I would’ve said – my pool name is rpool.

 

[pause]

 

I call the volume goldvol just to indicate that it’s the golden volume from which I create all the other volumes later on. Obviously, I’m not going to press “Enter” on that, but that would appear with absolutely zero space being occupied but appearing to be a 500 gigabyte emulated volume, the -s option being the sparse option.

 

[pause]

 

Now my Logical Domain has been sys-unconfig’d and it’s halted back to the OK prompt. So we’re sort of ready now to create the next cloned Logical Domain, which was the whole point of the webinar.

 

What I’m going to do shortly is take a snapshot. I went straightaway because I’m going to give you full details. I’m going to take a snapshot of that goldvol to preserve it and then I can take clones of the snapshot. The snapshot will remain completely [9:53 inaudible] in time, if you like, at that particular point. No changes will be made and I can in the future just clone as many copies of it as I need, each one thinking that it’s a 500 gig disk, but each one also are only using the actual space that it’s written to in the operating system. Sometimes that could be just a few megabytes so it’s very, very economical in terms of disk space.

 

Once I’ve cloned the disk that would be actually a writable volume and then I assign the clone to the next Logical Domain boot from it and then go through the configuration process, each clone being a basically a sys-unconfig operating system.

 

Copyright SkillBuilders.com 2017

×
Transcript

Guest Domains (continued)

Solaris Training Rapid Deployment of Logical Domains Part 4

 

>> Mick:  As far the initial operating system goes, I could carry on using it or I could convert it to use another disk and just leave the golden image there to use to create clones in the future. If I do need different versions of the operating system or radically different configurations then I would have to build Solaris on each Logical Domain separately.

 

[pause]

 

Here we are, going back a little bit creating the ldomvols zfs file system and then creating the gold volume. This actually creates within the operating system a couple devices that looked like this [00:51 inaudible] rather complex names unfortunately. Having found out what they are – but basically they emulate the path of the actual domains as I created the volumes as I’ve created them.

 

Here you can see the dev/zvol/dsk and then pod which is pool name where -rpool, ldomvols/goldvol. I can add that disk to my virtual disk service.

 

[pause]

 

So it’s as though I’m opening the disk tray, plugging another disk in which I named – I’ll do ldm add-vdsdev – add virtual disk device, the full path to the actual device name, which could be any disk path, by the way, and then name it golddisk@primary-vds0. Then we can start creating the guest domain which I did here that was seal.

 

Add domain, add a mathematical unit, virtual CPUs, memory – bear in mind, all these can be changed – a virtual network interface on my virtual switch called vnet0 and the disk. Add-vdisk with hpfdisk1 – just happens to be named after my domain name – and use that disk that I’ve just assigned to the virtual disk service and then assign that to the Logical Domain hpf file.

 

Say a couple of [2:43 inaudible] the boot device which will be hpfdisk1 and autoboot, which means the system would all automatically boot if the T-series server is rebooted. Not necessarily if the control domain reboots, but if the machine is powered off and then brought back again, not only will the control domain boots but so will this particular Logical Domain.

 

[pause]

 

>> Dave:  Mick, can I jump in with a question that’s in the queue?

 

>> Mick:  Certainly.

 

>> Dave:  Is a sparse volume the same as thin provisioning?

 

>> Mick:  Yes, it is, Dave. Yeah. Because of the fact that it doesn’t use any space, so as you know, in other circles, if you like, it would be known as thin provisioning. Every computer system has its own acronyms and terms. But yes, it’s the same.

 

>> Dave:  That’s great, Mick. I’d like to remind everybody keep your chat windows open. Thank you.

 

[pause]

 

>> Mick:  Having defined all the details for the domain like this, once we’re happy that all the resources have been assigned we could bind the domain and we can do a list-bindings just to check everything up. Then we can Telnet to the port and start the domain. You could do this in separate windows if you want to see exactly what’s going on or you could just do ldm start domain and when that comes back to the prompt do a telnet localhost in the port. Remember you can do an ldm list to find out what the port is, 5000.

 

Build the operating system. Here is could do a boot net because when I do a boot net I can find out what the MAC address is of my new Logical Domain and then I can set my Solaris build server to respond to that MAC address. Or if I don’t have that capability, I can actually create an ISO image, copy an ISO image of the operating system into my control domain. In this case, /iso/sol-10-u9, etc. That’s the downloadable image that I would get from Oracle. And add it as virtual disk service and then assign that again to the Logical Domain.

 

[pause]

 

When I do show-disks, it’s disk@1 that you can see when you do an ldm list like this to show your disk service. And I can boot and then I can say, “boot” and then that disk and then that would boot from the DVD and I could do my build server as normal.

 

[pause]

 

Here we go. And off we go, just a normal installation.

 

[pause]

 

A little bit of advice about you know making sure you choose the right MAC address. If you’re unsure about what listing is showing which MAC address, just do a boot net in your Logical Domain and then you can see what MAC address is requesting and then you can set up your build server 6:24 inaudible]. Or you can do it like I’ve just shown you and assign the ISO image.

 

[pause]

 

There we go, there’s a boot net. That particular Logical Domain is built in the normal Solaris fashion.

 

[pause]

 

What we’re going to do is we’ve built the Logical Domain, booted it, patched it, and then we did a sys-unconfig so we’re now back at the OK prompt. What we’re going to do, I’m going to stop the domain to a zfs list. There’s my goldvol which is the disk image that contains the sys-unconfig operating system.

 

[pause]

 

>> Dave:  Mick, I’d like to jump in with another question if I could.

 

>> Mick:  Certainly, Dave.

 

>> Dave:  The question is can I add resources to a guest domain later on?

 

>> Mick:  By all means, Dave, yeah. You can do it while the domain is switched off or you can do it while it’s actually running. Different versions of the software have provided different facilities.

 

For example, at one time, you couldn’t dynamically assign more memory. But now with the latest version you can even assign more memory. In fact, there’s a way of getting the system to adjust the resources itself. So the ldm system can work out but domain is not very busy and then another one is very busy and actually dynamically change resources for you. So the answer is yes.

 

>> Dave:  Fantastic. You know, Mick, I just like to say to our attendees here – there are a couple of attendees who responded to a question I put out, that they are indeed running Oracle database on Solaris. Oracle Standard Edition should be noted that is licensed by processor, no matter how many Logical Domains you’re running. Yet Oracle Enterprise Edition is licensed per Logical Domain. So it’s much more expensive typically. SE is very, very economical, standard edition, very, very economical way to go. If you’d like to know more about our experiences with the T-Series and Logical Domains with SE, just give us a shout.

 

Thanks, Mick, back to you.

 

>> Mick:  Thanks, Dave. So we have our goldvol that we’ve created with our Logical Domain which has now start and sys-unconfig’d. Remember that’s associated with an actual disk device effectively.

 

Copyright SkillBuilders.com 2017

×
Transcript

Guest Domains Golden Image, Snapshots,

Solaris Training Rapid Deployment of Logical Domains Part 5

 

>> Mick:  We then added a disk device as golddisk to our initial domain. Unfortunately, the commands do get a little bit convoluted and quite complex, so although the disk is known as one name to the actual virtual service, it’s then known as another name to each virtual domain and that hpfdisk1 which we set up as the boot device and then built the operating system on that together.

 

[pause]

 

Just out of interest, within the actual Logical Domain, the disk just shows up as a normal disk name, c0d0, etc. I hope that’s clear. It is a little bit hard to get your head around it when you first see it and you have to keep good records so that you can look back and think, “What was that I did two weeks ago?” Try and work things out.

 

[pause]

 

So we’ve patched it and stored it, rebooted to make sure the patches are applied properly and then sys-unconfig which is the stage you’re at now. Let’s go and create a new Logical Domain. Let’s clone.

 

First of all, we have to snapshot. And we snapshot goldvol. This will effectively create a read-only snapshot. The way that the snapshot syntax works is to say, “zfs snapshot,” the name of the volume and then @ and some character string like useme, for example. You can have your own conventions, date and time, whatever you like.

 

[pause]

 

There’s our snapshot. Now, zfs allows me to create a clone of the snapshot which is a rewritable copy.

 

[pause]

 

Zfs clone, the name of the snapshot, and then the name that I want to give it. Again, what shall I call it? I’m going to create a domain called sb1 and so I’m going to call this volume sb1vol.

 

I’ve effectively created another operating system. All I need to do now is to create my new Logical Domain and then assign that as a disk service to it. Remember this isn’t anything to do with the Logical Domain configuration. This just is opening out disk tray and letting in other disk to our virtual disk server.

 

[pause]

 

>> Dave:  Mick, there’s a question in the queue.

 

>> Mick:  Yeah.

 

>> Dave:  What’s a good storage solution?

 

>> Mick:  To go with this?

 

>> Dave:  Yeah. A good partner for a T-series server.

 

>> Mick:  I would think probably the Unified Storage, the 7000 Series because that’s being put forward as an ideal solution especially for something like Logical Domains and virtualization generally. With the capacities ranging from 20 terabytes up to three petabytes. Brilliant performance and nice and easy to manage. A T3 server and something like that would go very well together in this sort of technology.

 

>> Dave: Is that under the category of ZFS Appliance?

 

>> Mick:  Yes, it would be. Yeah, that’s exactly what it is.

 

>> Dave:  Excellent. Thanks, Mick.

 

>> Mick: Okay.

 

[pause]

 

Now as you can see, that’s already been done from a thing I did earlier. There we’ve already assigned the disk to the disk service. Apologies, the commands are slightly complicated. This is adding that device as an identified disk onto the disk controller, which I can then assign to my guest domain.

 

[pause]

 

So now I can go and create my guest domain.

 

[pause]

 

That’s the mathematical unit, eight virtual CPUs, three gigs of memory, and then I can do ldm add-vdisk. I’ll call it sb1rootdisk and this name up here.

 

[pause]

 

I forgot to put the domain name.

 

I hope that’s clear. I know it’s quite complicated the sequence of steps, but you do get used to it. So I’m adding a virtual disk which is going to be known sb1rootdisk to the sb1 domain using the disk that I’ve created just now from the clone.

 

Then I can do ldm list-bindings sb1, just to make sure everything looks okay.

 

[pause]

 

Then ldm bind sb1.

 

[pause]

 

Now you can see the new resources that have been assigned this time. I could also set variables like autoboot and so forth, but let’s get the domain started.

 

Let’s see what port we need to get the console on, 5001. It’s booting, as you can see.

 

We’ve created the operating system instantaneously. What we really have to do was defining the actual domain. There we go. and it’s going to come up now – because it’s a sys-unconfig operating system, it’s going to go into interactive mode and ask us to define all the particular features of the machine. Here we go.

 

You may be familiar with the step if you’ve ever done this yourself, similar to part of the stage of building the machine in the first place. One thing that springs to mind that you might ask is, “Is it possible to put a configuration for it to define some of this stuff?” If you’ve been used to using Solaris Jumpstart, you’d be talking about sysidcfg file.

 

The answer is no, not really. Although you can actually build one into your sys-unconfig image that does a partial configuration for you but configuring things when you do a first boot, the old trick of creating an image script. But you can’t really. You have to go through the actual machines basic configuration unfortunately.

 

[pause]

 

As I’m in the U.K., I’ll choose U.K. keyboard. But you’re probably all familiar with this. So here we are booting the machine and you can see it’s working fine other than creating the identity information.

 

So there we are. We’ve provisioned our domain very, very rapidly. All we have to do is complete the identity information.

 

[pause]

 

Here, in order to preserve the golden image, I don’t need the initial hpf1 domain anymore so I’m just stopping it. I’m binding it and getting rid of it. So I’m left just with the golden image that I can clone and there’s no danger of losing the domain and corrupting it in any way.

 

We have a couple of minutes left and I’ll carry on just configuring the machine just so you can see it’s all there working correctly. But this has been a real demonstration. I have done things for real. Nothing has been concocted, so you had a chance to see everything as it actually happens.

 

Copyright SkillBuilders.com 2017

×
Podcast
×
Free Online Registration Required

The tutorial session you want to view requires your registering with us.

It’s fast and easy, and totally FREE.

And best of all, once you are registered, you’ll also have access to all the other 100’s of FREE Video Tutorials we offer!

 

×