Transcript

Intro to VM Server (Solaris Logical Domains) and Core Multi-Threaded (CMT) Ultrasparc

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T-Series Servers Part 1

 

[music]

 

>> Mick:  Logical domains are basically a virtualization that’s on a specific version of the Sun’s SPARC processor known as the sun4v architecture and referred to as core multithreaded.

 

[pause]

 

You may have seen previous mentions of it as something called the core thread server. And over the past four or five years there have been a number of different processes available, in particular starting at T1 and the T2, T2+ and then the current T3 which is being superseded very shortly by the T4 which should start shipping in December.

 

These processors have a very large capacity in terms of the cores and the associated threads. So a typical core that we have a look at on the machine that I have here – the processor has eight cores, each of which has eight threads. The operating would see that as 64 different processors.

 

[pause]

 

There are ways of telling on a given T-series system exactly what physical resources are available. When it comes to creating domains, it’s quite straightforward to actually allocate resources between your different domain systems.

 

We’re using something called the T5120 to do this demonstration. Of course, this system also has other resources as such as memory, hard disk, PCI bus which can be split between various devices, and of course network interfaces as well as whatever things you might like to add like [2:17 inaudible] for example to connect to the SAN.

 

[pause]

 

This is what we’re going to look at during this demonstration. First of all, a little bit of positioning. Typical server uses and a little bit of a look at the server firmware which is a vital component of actually creating the domains and managing the systems, the firmware being the piece of software that is installed in the factory and sits between the hardware and the operating system.

 

The OpenBoot PROM, for example, in the POST (power-on self-test) software, and you’ll see how that works very shortly.

 

[pause]

 

Then we’ll quickly move on and look at the interesting aspects with practical demonstration of how you came about configuring the T-series server on installing Solaris although it does come preinstalled. Installing and obtaining the Logical Domain software, which is as Dave mentioned, now known as Oracle VM Server for SPARC.

 

[pause]

 

The types of Logical Domain that can be created. Creating something called the Control Domain which, if you like, is the controlling operating system from which you create and maintain the other domains which are typically known as guest domains. It’s within the guest domains that you create and run your applications.

 

We’ve pitched this webinar assuming that most of you will have had some Solaris admin experience. So I’m going to mention things like install Solaris but without actually explaining too much of how to do it and assume that you would have the skills from the past to understand what I’m saying.

 

Of course, there will be times when you can pitch in with some questions. Please do so at the appropriate times.

 

[pause]

 

As Dave mentioned also, SkillBuilders have been involved for quite some time now in selling, installing, configuring, maintaining T-series servers with some fairly complex configurations, one of which I’ll show you shortly to give you an idea of the level of complexity that you can reach.

 

[pause]

 

It’s a virtualization facility that allows you to divide hardware resources and run each system as a truly independent operating system. So it’s not like the zones or containers. It’s not a software feature. It relies on the firmware to create the logical divisions. You use the Logical Domain Manager software to instruct the firmware basically as to how to divide up the hardware resources.

 

It also has some fairly amazing facilities like being able to dynamically change the resources. In the latest version, you can dynamically change even the memory allocation of a system while the host is running. Not to mention assigning more CPUs if you think that’s necessary without having to do any reboots. So it’s quite an amazing technology. It sits somewhere between a software technology like Solaris containers or zones and something like the old SunFire Dynamic Domains that you may have come across in the past.

 

[pause]

 

You can run Solaris on a Logical Domain, of course. That’s the 11/06 release onwards. You can also run Linux and OpenBSD systems. There are a couple of links there in the notes for downloads if you’re interested. Obviously, the range of applications that you’d be running up the Linux being SPARC version would be slightly less that you would normally expect on a perhaps NIS servers. But nonetheless, you can run a Linux version.

 

[pause]

 

If you have any Legacy Solaris 8 or 9 systems, you cannot run them directly in a Logical Domain because they do not support the Sun4v architecture. But you can create zones via a container inside a Logical Domain and then you can run your Solaris 8 and 9 Legacy systems within that. And of course, with containers you have resource controls so you can assign a number of the CPUs of the system and the map of memory and so forth. So you can do some really good consolidations of an existing Solaris setup within your Logical Domain’s machine.

 

The machines have been referred to as a data center in a box. Take all your servers, put them into a T3 or T4 and then you’ve only got one physical system to worry about.

 

[pause]

 

Some of these are fairly obvious.

 

Oracle recommend small to medium size applications and databases. There are things like web financials, main services, departmental databases, good candidates, and development and testing environments. Remember that because you have physically separate systems with physically separate operating systems, you can patch them to different levels, which is very difficult or nearly impossible with a Solaris Container.

 

Of course, you have less physical hardware so restricting physical access can often be much more simple with something like one of these servers.

 

[pause]

 

The system firmware is a highly important part of a system like this and it’s called an ILOM (Integrated Lights Out Manager) and that maintains the components that you would normally be familiar with on the SPARC system like the power-on self test facilities and the OpenBoot PROM.

 

But basically what you do, you install Solaris on your T-series server and then you add the Logical Domains Manager. Using commands within that, you then establish the actual domains that are divided into the firmware.

 

[pause]

 

There are certain issues about versions of the Oracle VM Server software and its compatibility with the current version of the firmware. But when you download the software and you look at the associated read-me files, it will quite clearly explain the patch levels that you need and also the firmware revisions.

 

Copyright SkillBuilders.com 2017

×
Transcript

Demonstration: System Firmware, Network Configuration, Oracle Integrated Lights Out Manager, Solaris Installation

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T-Series Servers Part 2

 

>> Mick:  You can access the firmware initially usually through a serial port. I won’t go into too many details about how to do that. I’m sure some of you would already be familiar with that. Of course, once you’ve accessed that, you can then set up network connection and usually access through something like SSH, which is exactly what I’ve done here.

 

[pause]

 

This machine has got the operating system installed in it already.

 

[pause]

 

There you can see the description. I’ve installed my T5120 just like any other Solaris system so there’s no domain on there at the moment.

 

[pause]

 

The operating system version is, as you can see, update 9. I don’t know if you’re aware of an update 10 has just been released also.

 

In this particular case, I haven’t installed any patches or made any change in this. It’s just a straightforward install.

 

[pause]

 

The firmware side of things. It normally has DHCP enabled by default. So if you connect your system and switch it on while it’s connected to your network, it may pick up an address although it can be set statically.

 

The login defaults are root and the password is changeme, just in case you ever get stuck. There’s a handy bit of information for you.

 

[pause]

 

Here are some example firmware commands. Let me log into the firmware, which I can do with SSH because I’ve got the system set up. This would actually power on the machine, start/SYS. Then I could start the console.

 

When I do that, I would then see the OK prompt that you would all be familiar with from let’s call that a normal SPARC server. Once I’ve got that OK prompt, I can then proceed to build Solaris over from a DVD or maybe a network build.

 

These are more examples of setting network configuration within the firmware, which I can then commit to the firmware to make it permanent.

 

I can also examine settings. At the bottom there I’m actually committing the changes I’ve made so they are permanently now within the firmware and will survive at reboot.

 

[pause]

 

There are many more firmware commands and you can freely download or access the relevant manuals from Oracle. You don’t need to access it through My Oracle Support.

 

Just to show you, here I’m actually connected to the T5120, giving its IP address and it gives me a login prompt which root and then the password. Then we go.

 

[pause]

 

This one, you need it. Once more. Much better. There we go. A very fancy interface but very, very useful in this stage you having to know all the necessary firmware commands.

 

I had one warning it would stop there. That’s basically because I haven’t changed the factory default password, but everything else is looking fine in terms of the management of the system. A very sophisticated interface to look at on my server’s underlying faults or if I want to know the temperature of a fan or whatever. A very nice interface.

 

Even the ability to upgrade the firmware, assuming I have my Oracle support access to do so. There’s an example in the notes.

 

[pause]

 

Let’s have a look at the practical side of things now. Once I’ve got access to the firmware and though I’ve started the console, I would get a normal OK prompt as though I was on any normal SPARC machine, and I can then proceed to do an operating system install.

 

I won’t go into the procedures for doing so. What I would assume most of you have had some sort of Solaris installation experience. But of course you can setup a network server to do your boots which is exactly what I do because I’m forever performing builds in the training environment.

 

I would build Solaris in the normal way. The only thing I would say when you do this is that you have to make provision for disk space for your Logical Domains. That’s something we’ll be looking at a little bit later.

 

[pause]

 

Having installed Solaris, you then need to obtain the Oracle software – the VM Server SPARC software – and you definitely need a support contract from Oracle to get access to do this. For this particular file I’ve downloaded, it’s purely for development demo purposes as is the operating system itself. I’ve actually put this into a /root directory and I would now need to unzip it.

 

You see the resulting OVM server directory. That’s the readme file which will contain all the information about the firmware compatibility required for this version, the supported hardware, the patch levels and so on. Very clearly stated. That’s something important that you would normally want to look at.

 

[pause]

 

Again, to obtain patches and for firmware updates, you would need to go to My Oracle Support. If you’re familiar with Sun Support, there’s no longer the system handbook access and you have to do it through your patches and updates access on My Oracle Support, and then search for your server under the product or family advanced heading.

 

I put a little slide in here. Sorry, it’s not very clear. But basically, you have to search perhaps patches to find your firmware update and then you can download it and install it according to the instructions.

 

[pause]

 

Let’s just stop for a minute. Pause of breath. We’ve downloaded and unpacked the Logical Domains Manager software, which is here. We haven’t yet installed it. Theoretically, we’ve checked the readme file to make sure that we have the correct firmware version. We’ve downloaded that firmware as a patch from Oracle.

 

[pause]

 

That patch by the way for the firmware includes updates to the OpenBoot PROM and the POST software.

 

Now what we have to do is actually install the software. This is a real installation. I haven’t set up anything to emulate it. We’re doing it for real on this machine, so you’ll see precisely what goes on.

 

To check the patches – you’re probably already aware of this – you can use showrev -p to make sure your patches are up to date, and then you can crack the listing to find out whether the recommended patches are there. But in the interest of time, I’m not going to show how to install lots of patches.

 

Now we’re going to just do an install. While that’s going on, we’ll take a pause and maybe we’ll take a question or two if you have anything you want to ask. What I’ll do, I’ll just start the install.

 

Copyright SkillBuilders.com 2017

×
Transcript

FAQs and Demonstration: Installing LDOM Software, Creating Domains

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T-Series Servers Part 3

 

>> Dave:  Hi, Mick. This is Dave.

 

>> Mick:  Hi, Dave.

 

>> Dave:  A couple of questions in the queue. What command shows available processors on the T-series servers?

 

[pause]

 

>> Mick:  Thanks for that question. That’s exactly the same as any other Solaris system. The answer to that one is psrinfo, usually minus v to get a bit more information and stand back in amazement and not pressure its own.  

 

Here we can see the virtual processors starting at zero, operating at 1165 MHz in this particular case. The list just goes on and on and on and on, and up to 63, making 64. So the answer to your question, Dave, is psrinfo -v.

 

>> Dave:  Excellent. Another question in the queue, Mick. One of our administrators is just getting started to some extent. Where would be a good place to find out how to install Solaris? Where would this person get started?

 

>> Mick:  The best place – you know what I’m going to say, Dave, obviously – is come on a training course. There are many different types of machine. There’s Intel processors. You know this type of system. There are the non-Sun for the M-series. Each of which can have Solaris installed on it but in a number of different ways.

 

Of course, there’s plenty of documentation in the Oracle website to help you. It can take a little bit of time to work out a particular procedure. My answer would be come on a training course and that is the best way. You couldn’t really distill it down into a short list that it could apply to any particular server.

 

[pause]

 

>> Dave:  I’ve just put in the link for the group. A link to one of SkillBuilders administration classes so that might be helpful for one or more of the students.

 

>> Mick:  Great. One thing I would say is once you’ve done a Solaris install like any operating installation that you ever done, it always follows the same path. How do I partition the disk? What software options do I want? What’s the date and time? What’s the time zone? All that is a standard on any install. But like now when we do an install, we have to make other considerations like the amount of disperse.

 

Let me get this software installed. Unzip the software distribution, CD to the directory. I’m running the install. It’s asking me, do I want the Configuration Assistant which supposedly use a friendly frontend to configure Logical Domains, but I’m one of these guys – like many who’d like to see exactly what’s going on. So I’m going to answer no.

 

[pause]

 

Now Logical Domains is installed. There are two packages that were installed, one of them being the LDOM software. Another one that we’re not going to be using that allows you to convert physical machine images into Logical Domain.

 

There’s a little transcript of the install.

 

[pause]

 

There’s a dmd running, the background process which does all the work for you. When you use the ldm command, it communicates with the dmd, which is quite the typical way of doing things on Unix or Solaris box. In this case, it’s called ldmd. We better make sure that it’s running. There we can see that it is enabled, it’s online, it’s looking good.

 

We can also now use the ldm command which is installed as part of that vm server for SPARC software. The first thing I’ll do is enable the vm list which shows what domains I’ve got. It’s showing my machine. There’s the primary domain. This is the control domain. This is the machine that has control over everything we’re going to do and it has all the resources available to it, 16 gigs of RAM, 64 virtual CPUs which we’re going to change fairly rapidly.

 

We could actually run the machine just as a normal Solaris box with a heck of a lot of virtual CPUs and memory, but we’re not going to. We’re going to create some domains. First of all, we’ve got a control domain, which is the original Solaris system that we’ve just been looking at. We’re also going to have something called a service domain which has virtual services associated with it like a network or a disk controller.

 

We can also create a I/O domain which is another input/output domain but actually has direct control over some of the hardware. So we can install the domain and then maybe portion part of a PCI bus or a complete PCI bus to that domain. This is something we’ve done on a system to allow us to do I/O multipathing. So the system has two host bus adapters and one domain for those, the other one failover and all the domains carry on running. It’s also used where we split the gigabit network interfaces between the domains.

 

Having created those domains you then create your guest domain. In practice, on a smaller sort of less capable T-series server, your control domain will normally carry out all three functions and indeed the hardware might have not have the capability of being split so you can’t really create a separate I/O domain for failover. But certainly something like a T3, it’s evidently suitable to do that sort of thing. We’re just going to configure the control domain and we’re going to associate all the services with that and then we can look at creating a guest domain.

 

[pause]

 

First thing we have to do is set up virtual services. We have to create a disk controller. We have to create a network controller to which we can then assign devices that are going to be used within our virtual machines.

 

Bear in mind, we’re doing a real installation here. We can go through these steps so the system doesn’t currently have any of these things set up. You see me having to type it all in.

 

[pause]

 

This is creating a disk controller in the primary domain and the add-vds is understood as meaning create a virtual disk service which I’m calling primary-virtual disk service 0 because I can have more than one.

 

Then I’m going to create what’s called virtual console controller which will allow me from the control domain through telnet into my Logical Domains to display the [8:46 inaudible], for example, as though I’m on the system console. Then I’m assigning the port range locally through telnet to see how to this. You’ll see how this works a little bit later.

 

[pause]

 

Here I’m adding a virtual switch. There are some very powerful facilities within this software. This is actually creating for all intents and purposes a network switch within the control domain. Once I’ve got that switch, I can then assign network ports to my guest domains. In fact, I can have more than switch through different I/O domains and then I can use IPMP (Internet Protocol Multipathing) to provide failover across the different I/O domains, which is something else that we did on the T3 for I/O.

 

[pause]

 

If I want to look at the services I’ve assigned at any time, I can do ldm list services.

 

[pause]

 

Not quite as straightforward as in the Unix books. We’ll have to copy and then paste. There we go. The reporting facilities of ldm are quite good.

 

Copyright SkillBuilders.com 2017

×
Transcript

Demonstration: Configuring the Control Domain

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T Series Servers Part 4

 

>> Mick:  Now we have to configure the control domain and allocate resources. There are three main types of resource. The processes which we can think of in terms cores or threads – there’s one mathematical unit associated with each core and on the amount of memory. Because the control domain isn’t going to be doing an awful lot and it’s not going to be running applications, I don’t have to be too generous.

 

I’m going to provide it with one core, one associated mathematical unit and 2 gigs of RAM. Currently, the system has everything – 16 gigs, 64 CPUs.

 

[pause]

 

I’ll allocate the resources.

 

[pause]

 

I need to find out what it’s currently got which I can do with ldm list bindings, which is a slightly more verbose version of doing the ldm list. I’d better type in there more. You can see all the core ID and thread numbers, all the virtual CPUs, and the memory. The virtual controllers that I’ve set up just now as well showed up.

 

It’s a good idea if you’re going to do quite a lot of configuration to do list bindings and save it in a file. Just in case you’re in a hurry, you need to remember what the original resources were that you started with. I think a good idea all the way through is to keep good records of the configuration because it can get quite complex.

 

[pause]

 

I can list individual resources as an example. The MAUs are also known as crypto units so ldm list -o crypto and the name of the domain. That’s more relevant if we had guest domains then we can use the domain names to find out what resources are allocated against a particular Logical Domain.

 

Now we’re going to allocate the resources. In a minute we’ll be doing a reboot and that’ll be another opportunity to ask questions.

 

[pause]

 

We’re in the process, copy, paste.

 

[pause]

 

I’m going to set a number of CPUs.

 

[pause]

 

Usually, it’s done on a per core basis, certainly we guest domains, as the recommended way of making things the most efficient, but not too vital with things like a primary domain or indeed an I/O domain.

 

What I’m doing here – sometimes the system is confused if I start messing around with the memory. So rather than try to do a dynamic reconfiguration of the control domain memory right now, I’m going to get it to do the reconfiguration a bit delayed until I do a reboot. That’s the start of reconf.

 

Now I’ll set the memory but this won’t take effect until I actually do the reboot.

 

[pause]

 

Now, I’ve got everything set up. I need to do a reboot to make all those things come into effect. I’m just going to do a straightforward reboot. Don’t forget if you’ve got applied patches on the whatever, you would normally do it [4:37 inaudible] 6 to make sure they get right. So I’m going to do a reboot and then we can pause for a minute or two and take questions.

 

The server of course is now gone away until we access it again.

 

[pause]

 

Let my close my PuTTY and hand over to Dave to see if he’s got any questions.

 

>> Dave:  Thanks Mick. Indeed we do. A couple of questions in the queue.

 

[pause]

 

First one, I believe I can handle. Is the presentation we’re doing today going to be available afterwards?

 

Yes, it will. We record these presentations and put a link up on the website for them. We will send all the attendees the link to the recorded presentation so you could view it any time you want.

 

Second question, Mick, you’ll have to handle it for me.

 

>> Mick:  Sure.

 

>> Dave:  What if you want to know the allocated resources for a specific LDOM?

 

>> Mick:  Right. I could show this a little bit in more detail later. If we go back – you can do ldm list-bindings (space) and then the domain name. So we could say ldm list-bindings primary.

 

Very shortly once the machine reboots, we can have a look at creating a guest domain. As part of that process, I will give the guest domain a name like dbserver or something. Then I could do ldm list-bindings dbserver. That’s the answer to the question, Dave.

 

>> Dave:  Great. One more question, Mick.

 

The question is estimated duration. How long do you estimate it takes to set up and configure these guest domains?

 

>> Mick:  The initial one has to be installed like any other Solaris system. A couple of hours, if you’ve done or planned it properly and installed it. But you can install it in such a way especially using the ZFS file system – where having installed it, you can do something called the sys-unconfig which takes away all the identity of it, and then you halt it. Then you can actually snapshot the entire operating system and clone it. You can incredibly roll out an entire operating system in a few minutes.

 

[pause]

 

>> Dave:  Great.

 

>> Mick:  Which I did with [7:43 inaudible].

 

>> Dave:  Pardon, Mick. I think I interrupted. Say that again.

 

>> Mick:  Which is exactly the way went about it with [7:51 inaudible].

 

>> Dave:  Very nice.

 

>> Mick:  But of course, and if you want to, if the operating systems are widely different in requirements terms, you can install each one individually. On that though, you’d be looking at a couple of hours on each case.

 

>> Dave:  Thank you, Mick.

 

>> Mick:  You can roll them out in minutes.

 

[pause]

 

My PuTTY.

 

[pause]

 

Dolphin is the firmware, by the way, and whale is the logical domains control domain that we just rebooted.

 

Copyright SkillBuilders.com 2017

×
Transcript

Demonstration: Configuring the Control Domain (continued) and Using ZFS Disk Space

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T-Series Servers Part 5

 

>> Mick:  Thankfully, for our demo purposes, the machine has come back fine.

 

[pause]

 

There’s our ldm list, if you remember at 64 CPUs and 16 gig of RAM, now it’s only got 8 VCPUs and 2 gigs of memory. All the other resources are now available to us using ldm commands to configure guest domains so that the control domain is now up and running.

 

[pause]

 

There are issues about when you use an ldm command. It stores information within the firmware, but there’s an intermediate phase where it’s recorded within the operating system that which you have to be aware of and make sure that information is correctly written to firmware.

 

There are various commands like to add configuration to the firmware. That would take whatever the current configuration is and write it away with the name called primary-initial. The only one list by default is something called factory-default.

 

So if you’re experimenting and you make a mess then you want to clear out what you’ve done. There are a couple of ways of doing it. You can do it from the operating system. You can set config factory-default or you can do it at the ALOM, the Lights Out Manager, about doing this command at the bottom of the page here.

 

The one above would be on an older T1000 or T2000. I’m not going to bother doing anything about that because I’m purely demonstrating.

 

One thing we now need to do is make the virtual switch we created accessible to our physical network which currently isn’t. Then that way any guest domains that have virtual network interfaces connected to that switch can then access the outside world. If you want the complete data center in a box without any access to the outside world, you can do it without doing the next step.

 

[pause]

 

Let’s find out what network interfaces do I have.

 

[pause]

 

It takes a little bit of time to come back. There’s my virtual switch. What I’m going to do is configure that as my main network interface. I normally would but in the interest of time, I’ll just talk you through it.

 

Currently, I’ve got an e1000g0 interface through which I’m communicating. What I would do is replace that with exactly the same configuration but for vsw0 interface which is shown. The procedure for doing this is shown on this page here, 32.

 

[pause]

 

Take down the e1000g0, unplumb it which effectively if you like removes the driver from kernel, plumb the vsw0 interface instead and bring it up. Then create the necessary file in etc/hostname.vsw0 rather than hostname.e1000g0. The next time we reboot, it will use that interface. In that way, the switch now has access to the physical network.

 

[pause]

 

We can actually investigate properties of the network settings rather than list-bindings and displaying everything. The ldm command has some pretty good ways of displaying specific information. It’s just trying to remember what the commands are. But there’s an example to look at the network.

 

When we set up networks within our guest domains, there’ll be vnet0, vnet1 and so forth, and we can configure them like normal physical interfaces.

 

[pause]

 

One thing we have to do also is enable this vntsd daemon which will them give us access to the [4:43 inaudible] ports so we can access the OK prompts of any logical domains we create.

 

[pause]

 

When we install guest domains, we will provide resources to the domains and also disk tiers. The disk can be a physical hard drive. I created the [5:13 inaudible] file or it can be a ZFS disk, which is what we’re going to use.

 

This system is using the ZFS disk, which you can then see there’s a df listing. If you’ve never seen ZFS before, it looks a little bit different than normal.

 

[pause]

 

First of all, to find out what resources we have available, you can do an ldm list – devices and that will show the actual hardware that’s currently available to all the systems.

 

[pause]

 

I’ve explained that you can use a number of different disk backend and you can also extend a domain. You can of course use SAN devices, ISCSI, whatever you like, just like you can for any system. We have a ZFS file system and what we’re going to do is do a ZFS list. We have a pool called rpool.

 

[pause]

 

What I’m going to do is create a little dataset within it to store our Logical Domains volumes. ZFS has this nice little facility of being able to clone system. If we create a volume within ZFS, we can then snapshot it and clone and automatically be able to roll out another copy of the operating system. If we sys-unconfig before we halt it, when the system boots, it will go through an identification process asking us the host name, the IP address, and so forth. Every time we want to create a domain we simply clone the snapshot, assign it to the domain. We put it, and answer the questions.

 

We’ve got ourselves a brand new operating system that takes up very little space, so it’s very economical at being very quick. It’s not always the best way to do things, but certainly if you’re going to roll out a lot of similar Logical Domains, it’s a great way of doing it.

 

[pause]

 

It’s often referred to as the “golden image” as it’s shown on the slide here. Let’s have a look.

 

First of all, we got to create the holding area, if you like, the dataset that’s going to hold our volumes. The pool on this example is called pod but mine is called rpool. Now I’m going to create an emulated volume.

 

[pause]

 

This is by the way – sorry, it’s [8:37 inaudible] these terms around, emulated volume. But what I’m effectively doing here is creating a disk device sitting on ZFS that will look like…

 

[pause]

 

Where are we? We’ll call it goldvol. We’re going to use it as an image. But we’re going to partially just going to create a domain using that to search the disk. That will create actually associated disk device names that we can now use and assign like the example there at the bottom. So we can do ldm add-vdsdev /dev/zvol/dsk/rpool/ldomvols/goldvol golddisk@primary-vds0.

 

Remember primary-vds0 is our disk controller so we’ve attached a disk to our disk controller and we’ll use that very shortly. Let’s quickly go through adding a domain.

 

Copyright SkillBuilders.com 2017

×
Transcript

Demonstration: Creating the Guest Domain

Installing Oracle VM Server formerly Solaris Logical Domains on SPARC T-Series Servers Part 6

 

>> Mick:  Another five minutes and we should be pretty well there.

 

[pause]

 

I’m just going to use my copy and pasting again.

 

[pause]

 

I’m going to call the domain hpf1. It can be called anything. What will happen it will replace the initials of hpf.

 

Here I’m adding the mau unit, adding the CPUs. This is a repetitive process and the only thing that varies between different domains is the actual resources you configure which can dynamically be changed while you’re running. So if you can’t quite get it right, don’t panic. You can add or subtract any of these resources dynamically.

 

Here I’m adding a virtual network to vnet0 so that we’ll have a vnet0 that we can configure with ifconfig when the domain runs. Here we are assigning the disk, making sure the domain boots. We just copy that and set the default boot device which is the disk that we’ve just assigned.

 

[pause]

 

And then fingers crossed, we combine the domain which then assigns all the resources. It would’ve given us an error had I made any mistakes with all that typing. That was the most tricky part of the presentation. Again, as I’ve shown you, you can do list-bindings to find out what those bindings actually are.

 

[pause]

 

I can do a list domain and there’s our list-bindings which shows you the number of resources that we’ve actually allocated. Then I can write my configuration away. Then I can start the domain. I’ll do it the other way around.

 

[pause]

 

If I do an ldm list, I can see the port to which I can telnet to get the console. I say, “Telnet localhost.” Don’t forget that this is in the control domain to that particular… And there we go. We’ve got our new Logical Domain and there’s the OK prompt which responds to all the normal SPARC OpenBoot PROMP commands that you’re familiar with.

 

Now we can do a boot net. We might have to set up a boot support on a boot server. We can associate by the way the same way that we associated the actual disk device, we can associate a CD or ISO image to build from as well.

 

[pause]

 

>> Dave:  Hey Mick, we’ve got a few questions in the queue, if I can throw them out to you.

 

>> Mick:  Yeah. We’re just about finished actually, Dave. I’ll just put the last slide up.

 

>> Dave:  Perfect. I’ll hold them. Thank you.

 

>> Mick:  Just to remind people about the SkillBuilders capability in this area. But thanks, everybody, for listening. Let’s have some questions.

 

>> Dave:  Thanks, Mick. A couple of questions straight away. Can you recap the difference between a Logical Domain and a Container, maybe benefit, disadvantage, or pros and cons?

 

>> Mick:  The Logical Domain is a distinct hardware feature. You see you are effectively creating the physical system and you install and maintain the operating system within that or separately. Your admin load will increase just a little bit more the Logical Domains, but you could have completely different operating system instance, patches, completely different levels.

 

With the Container, that is a software facility. So if you had the Solaris box with Containers, you’d be running an initial version of the operating system called the global zone, and then you would create further zones or Containers within that, but they’d be using the same [5:18 inaudible].

 

So the virtualization is done through the kernel facilities and other daemons that maintain the separation between the different zones. It’s virtually impossible, for example, to have those machines running different patch levels certainly within the OS.

 

That’s the free technology that comes with any Solaris box and highly suitable for certain types of application. But when you want full separation with different hardware, for extra stability where no Container can impact on the other, then a Logical Domain would be the obvious choice.

 

[pause]

 

>> Dave:  That’s great. Another question. Can I use SSH to access the LDom?

 

>> Mick:  You can, Dave, but not until the LDom is actually installed with Solaris. You can’t access the OK prompt like I’m doing using SSH. You would only be able to do that with the firmware.

 

To get access to the OK prompt with the Logical Domain, I would have to log in to the control domain first and then use telnet. But once the Logical Domain is built and you’ve configured SSH maybe through access, then you can log into it just like any other system. That’s the answer to that one, Dave.

 

[pause]

 

>> Dave:  Great. Mick, can you also recap the critical domains, for example, I/O domain? There’s also control domain and guest domain. What are the memory requirements associated with these domains?

 

>> Mick:  They are service domains, if you like. But the main domain is your guest domains. So if you took that as an example with the T3, their guest domains are running in Oracle databases and therefore each domain is sized according to requirements of the applications you guys run, just like you would size any other Solaris system.

 

The control and the I/O domains are maintained separately. Although they can run applications – in the particular case of the client we know about, they are purely there to service the guest domains and therefore they only need enough resources to run the operating system. A single core is absolutely fine and a couple of gigs of RAM. But just like any other domain, if we find that that allocation is not quite enough, we can dynamically change it while the machine is running.

 

Now if we go back through the notes, I can show you – going back to page 35. In my current project, we’ve created a control and an I/O domain so we created a Logical Domain and we’ve given ownership to the PCI bus to it, so we can actually physically apportion the hardware to an I/O domain. So the control domain owns the PCI bus and the I/O domain owns another PCI bus, each of which has an HBA controller connected to the SAN. The guest domains are configured so that they will use one or the other hardware, and if one of those fails it will switch over to the other domain.

 

Having said that, not all T-Series have the ability to split the PCI buses which is so [9:17 inaudible] the T3 server that we’re working does. That can also be done by splitting. You can split the network interfaces across both domains also and you can therefore create multipathing such that if one domain fails, it will automatically use the other.

 

So it’s surprising the level of resilience you can build in to a single physical machine. In our case here, the control domain acts as a single I/O domain and the service domain as well. The service domain has the virtual services associated with it rather than the physical ones.

 

[music]

 

Copyright SkillBuilders.com 2017

×
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!

 

×
Podcast
×