Well, a few weeks have gone by since I made my little announcement. I’m happy to say things are going well on the creative front, and I’ve been toying with the idea of recording some songs and putting them on her for your delight and delectation. This post is about my computer life. As promised I’ve started contributing to the VMUG Wiki project. I must admit the first couple of attempts were disaster. I had technical problems as my home lab has been down for months, and the process of updating the vSphere5 content to vSphere6 content is actually – sorry to say this – tedious.
Anyway, last week and the week before – I got over my technical problems and started to find an approach for updating the content that feels quicker, and therefore less tedious. What floats my boat is writing new content about something that has changed significantly, rather than playing a ‘spot the difference” game with software that if you blinked you’d miss any changes. Anyway, someone had to do this job, and as I have time on my hands now, it might as well be me.
So I’m pleased to report we have a new chapter on the VMUG Wiki (probably the first in months!) called thrillingly “Installing VMware ESX 6“. Yes, I know there’s a broken image in the content. I’m working to resolve that – we have a more serious problem with images in Wiki generally – as MediaWiki sometimes fails to create “thumbnails” correctly. I’m working with the VMUG folks to try and resolve that.
For those like me who have been in this game for a while – there are no surprises here. Indeed some of the content is so similar I haven’t bother to swap graphics around when the only difference in the UI is the bloody build number! However, there were one or two new things that caught my eye.
Firstly, “Lockdown Mode” (that highly popular feature of ESX 😉 ) has a new option or mode called “strict”. I understand this prohibs the use of the DCUI to turn it off. So the only way to turn off “Lockdown Mode” is if the host is manageable via vCenter. That makes vCenter the only method by which the host can be managed. I guess this removes a ‘backdoor’ method caused by the root account being compromised. In my experience customers (except govt/military types) rarely use this feature – in fact many people lower security by enabling SSH which is normally disabled.
Secondly, the other thing I noticed is if you attach two vmnic to vSwitch0 they marked as being Active/Active. Previously, ESX marked one as being Active (vmnic0) and the other being Passive (vmnic1 for example). I consider this an ‘improvement’ but I imagine most experience VMware shops pretty much have the network setup nailed down by now and automated – and never use this method anyway.
Thirdly, I notice the TCP configuration has changed slightly for DNS (this might have changed a while ago, and I hadn’t noticed). You can now set a IPv6 DNS address (whoopee!), But the ‘odd’ thing is the option to set a secondary DNS for IPv4 seems to have disappeared. I assumed you could use comma separated values in the box – but it doesn’t seem to accept that. I dunno if that’s “by design” or bug…. If someone in the ESX team is reading this and knows their onions I’d be interest to know what the rationale is behind this… so I can feel less stupid and educate the community.
Finally, I am working on Chapter 2: Installing vCenter. That’s proved to be a bit more interesting given that there’s the new “Platform Service Controller” (great sexy name there!) and new Postgres support for the Windows version of vCenter. I increasingly find setting up the Windows vCenter a total ball-ache. That’s not because of VMware, but because of the bloody Windows dependencies that must met first. I mean why does anyone bother with the Windows vCenter for new deployments (lights touch paper and walks away!
Note: If you have never seen this parody of slone ranger types on their “Gap Yah”, you have missed out…
Today was officially my last day with VMware. I’ve been at VMware since August 2012, initially securing a role in the competition team as the “Senior Cloud Infrastructure Evangelist”, and then I moved into a “Senior Product Integration Architect” role in the EVO:RAIL Team in August, 2014. I was at VMware for 3-years, but of course it feels much longer, because really I’ve been solely focused on VMware technologies since 2003. This year was my 10th VMworld, and I’ve attend both the US and EU events for 10 years in row. That’s 20 VMworlds (for those who find 10×2 a bit of challenge).
It’s been crazy 3-years, going from essentially company of 1 person as independent freelance contractor – to joining a company of 20K strong with billion dollar profit was quite a transition. One I hope I successfully executed on, and I would cheerfully recommend to anyone in similar position that you spend sometime in “VendorLand” in your career. You owe it to yourself to be exposed to as many different perspectives and experiences in life. And that was partly the reason I joined the company. And if you like that also a reason for moving along.
I’ve been in the IT industry since 1993, and whilst I’ve had some ‘breaks’ such as doing Masters in American Studies in 1996, and taking 3-months out to travel round the US in 2000 – it has been a long time since I’ve taken time out to do something that’s 100% about me and my interests. The last 10 years have probably been the most significant to my career – for my own satisfaction I thought I might be interesting to list those achievements of the last 10 years:
Became one of the first freelance VMware Certified Instructors (VCI) in Europe
Pivoted the “RTFM Education” website to be one of the top go-to locations of quality content on virtualization
Sold the said website to a media company in Boston, MA
Spoken at practically every VMworld event since 2006.
Spoken at innumerable VMUG UseCon events in the US and EU, being state-side one-week every month for two years..
Authored 9 books on VMware technologies – two of them self-published with money donated to charity
Launched and ran two successful podcast channels for five years – the Chinwag and Vendorwag
Co-hosted the VMware Communities Podcast with the industry legend that is the mighty, John Troyer…
Raised money for UNICEF and others charities via book royalties and the successful “VMworld Swagbag” Competition
It’s my guess that in 12-months time when I’ve recharged the old batteries and little grey cells – I will need to come back this bulleted list as reminder what I have achieved and contributed to the industry. I personally feel I’ve lived through an exceptional period in our industry. 10 years ago or more there was no such thing as virtualization in production x86 environments. VMware cut swath through the datacenter, and radically changed the way do things – and continues to do so. I’m quietly proud that in my own small way I had my own part in that story….
So it feels ‘right’ at this stage in my life to step back and take sometime out. After all we only have one life, and its not always good idea to defer things into some hazy future that never arrives. So I’m taking what I call an “Adult Gap Year”. Of course “Adult Gap Year” is a bit of joke, on the way the current generation seems to take a break before or after university to go ‘travelling” and decompress after all the pressure of high education (yes, its so tough having 10 hours of lectures a week, right?). But in similar way I kind of feel a break would do me the power of good.
So what am I going to do in the next year? Put simply all the things I’ve been saying I was going to do in the last 10. I’ve had ideas and ambitions of non-IT nature that I’ve been thinking about for ages. The vast majority (in fact all of them, and that’s some majority) are of a creative format….
Committed To Community: It’s my intention to carry on speaking at VMUGs. In fact in recent years I spoken less at VMUGs because my responsibilities to the company, projects and customers had to come first. So I’m hoping to spend sometime state-side speaking at UserCons. The VMUG community is something I feel passionate about, and it will be my way of staying connected with y’all even whilst I’m taking my Adult Gap Year. I’ve decided to spend at least 1-day or perhaps 2, supporting two key initiatives – the VMUG Wiki and Feed4ward projects which I helped kickstart. I really care about these two projects, and I’ve begun to realise whilst its one thing to help launch these things, the real ‘graft’ is in the continued support and development.
Travelling Man: Like many a road warrior I’ve seen an awful lot of airport carparks, airport terminals, taxis, hotel rooms, and business parks and convention centres. I rarely get enough to time explore places. I have more than a tourists interest in the United States, having done a degree and post-graduate research on its literature, cultural and history. During my research its become clear that I want to discover and explorer the national parks of the US – as well as my own area of Derbyshire and Peak District. Along side I want to write a Travelog/Reportage/New Journalism account of the journey, which I hope to self-publish in a book form, hopefully with photographs of the places I visit.
Poetry Please: I’m poet. And I know it. Hope I don’t blow it. To quote Bob Dylan for second. In my teens and early 20s along side writing lyrics for songs, I also tried my hand at some poetry. That’s a passion I recently re-discovered when I moved to the country. I joined a local poetry group and rekindled the interest. I have ideas for two collections of poetry – and have a working title for the first called “False Confessions” (The title is reference to the idea that you cannot trust confessions that are made under-duress, something our friends in Bagram and GitMo never really understood). The working title for the other collection is “False Memory Syndrome”. The theme is about memory and how much humanity can trust memory and history. I hope to self-publish these two collections by the end of next year…
I could make a wild sensation as rock ‘n’ roll star: As some of you might know I’ve got a big passion of music. This year i joined my local song writers group, and penned my first song in about 25 years. Right now, it doesn’t feel like I’m going to have the creative head-space to write poetry and songs (and yes, I do know there’s some cross over in both directions). But what I do want to do is be more out there from a performance perspective. Currently once a month I walk down to my local pub, and hammer out about 3-4 songs along with my fellow musicians at our local “acoustic session. There’s a very healthy live music scene in my area, and I want to get on that circuit – and build up my confidence in performing to a crowd. I think the way to be a good performer is to perform frequently – and once a month isn’t really cutting it. Plus I tend to bring new songs to the group every month (after all you can’t trot out the same song to the same people every month!) But that means I rarely perform the same song more than once. I think the way to get better at performing isn’t just practise and rehearsing – but performing that same material multiple times to different audiences in different venues. At the moment I’m thinking of focusing on the local pubs ‘session’ nights, but I’d like to try my hand at the bigger “open mic” slots in the larger towns and cities near where I live.
Oh, for those who don’t get the reference (sigh…)
So yes, I know a long blogpost. Is there any other from Laverick? But I wanted to explain my thinking and the rationale about taking this time away. I hope to see you all in my travels….
I’ve just registered for the first London VMUG of next year kicking off the season on the 21st Jan. I won’t be speaking, so I will be just another attendee in the crowd.
There’s the usual round up of presentations from VendorLand including Bitdefender, Velostrata and Tegile – as well as VMware themselves…
The start of the new year will be chance to have heated panel debate as well as 2015 Community Speaker Awards… and the all important vBeers after the meeting sponsored by 10Zig…
This year meeting will be opening a new chapter on the London VMUG. After this years finale of the UK National Event much of the steering committee from the London User Group have opted to exit stage right on high. That’s sensibility that resonates with me – wanting draw things to a close – not with whimper but a bang… So the 21st will see an introduction to a brand new steering committee and I think is chance for “Fresh Blood” to come into the group – whilst at the same time giving ‘respect is due’ to the former members for all their hard work in the past.
A couple of weeks ago the national VMUG event was held in Birmingham. As ever I was there with the VMworld Swagbag and supported by Michael Poor and Barry Coombs in the process of raising money for this years good cause. This years good cause is “CodeClub”. They are a not-for-profit that run after school coding classes in the UK for children between the ages of 9-11. I’m pleased to say we managed to raise £500 for their worthy organisation…
Well, would you believe where the time has gone? Another year and another UK VMUG beckons as well as new SwagBag Competition. In case you don’t know – for the last couple of years I’ve ‘bagged” a VMworld Bag, and stuff it with quality ‘swag’ gathered over the year. Each year this bag is raffled off at the UK VMUG event held in the November. The money raised is donated to good cause. This years good cause is “Code Club” (https://www.codeclub.org.uk/) it’s goal is to help 9-11 learn the first principles of programming – rather than just being users of Word, Excel and the Internet.
To stand ANY chance of winning the Swagbag you must attend the VMUG vCurry event or the UK VMUG event itself. The UK VMUG event is held on Thursday 19 November 2015 – at National Motorcycle Museum, Solihull. The vCurry Event happens in the same venue the night before, and usually incorporates a quiz. The UK Event has special guest visitors including Josh Atwell and the know legend in his own lifetime, John Troyer.
Full Details for the UK VMUG and vCurry Event are here – Register Today!
Anyway, that’s it from me – lets have a look at the bag and this years award winners…. I call it the Oscars for Swag!
UPDATE: Oh, I forgot to mention to other additions to the bag. Firstly, I’ve got one of those “Tile” things to give away. It was vExpert gift. Basically, you stick a ’tile’ on something important that you frequently lose (like your keys for instance), and your phone will locate it with a special app. Sadly, you cannot use a tile to find your phone (which for me is more common…). Ravello has also offerred a free subscription to their service – in case you don’t know Ravello allows you to run nested ESX in the Amazon EC2 cloud – which could be the next incarnation of the homelab. Finally, PluralSight have included a free subscription to their training. You might know PluralSight acquired TrainSignal a while back which was the go-to source for training on VMware technologies.
One of the big differences in how vSphere works as deployed by EVO:RAIL is with DNS. As you might know vSphere has many requirements for name resolution, and often various vSphere features will not function or setup correctly without DNS being available. A classic example is simply opening a Remote Console window on a VM. Although that request might be triggered from a vCenter session, it’s actually the VMware ESXi host that handles the redirection of the video, and allows for Keyboard, Mouse and Screen (KMS) functionality. Remote Console sessions require name resolution to the VMware ESXI host to work. I could go on at length with other examples but you get the picture.
The good news is the EVO:RAIL Appliance takes care of all these requirements. In fact EVO:RAIL has its own built in DNS service. This means that there are no service dependencies required to setup the appliance at a green-field location. That’s right, the EVO:RAIL appliance will configure itself – even if there’s no DNS, DHCP or Active Directory.
This does mean that the way name resolution is achieved is different from standard vSphere as deployed manually by customers. With vSphere the path of the name resolution from the VMware ESXi host is via its management network. For example after installing VMware ESXi the customer assign a static IP address and configures the VMware ESXI host for its Primary and Secondary DNS, as well as its domain suffix, using something like the Direct Console User Interface.
In this case the VMware ESXI host queries the corporate DNS server directly. With EVO:RAIL this behaviour is similar, but different. The EVO:RAIL Configuration Engine will set static IP addresses for the ESXi management network and also set the preferred DNS settings – however, what is queried is the built-in DNS Server of the EVO:RAIL.
So in this case the DNS query takes this path:
ESX Host >> vCenter Server Appliance DNS Service >> If not internally resolved forward it is forwarded on to the corporate DNS server.
You can tell that a DNS service is running with the vCenter Server Appliance using the command “netstat –natlp | grep ‘:53’. As you might know all DNS queries are responded to by listening on TCP port 53. This will show that there is a “dnsmasq” service running.
The dnsmasq service holds hostname records in a text file on the vCenter Server Appliance /var/lib/vmware-marvin/dnsmasq/hosts. Usually, this will contain at least the four VMware ESXI hosts that make up the EVO:RAIL Appliance together with the IP address and FQDNs for the vCenter Server Appliance. In the new release of EVO:RAIL we will have a dedicated virtual appliance for managing the physical appliance that we are calling the “EVO:RAIL Orchestration Appliance”. You can see it listed in the screen grab as evo04-evorail.vsphere.local.
If you add a second appliance to double your compute and storage resources the hosts file would be updated to include those FQDNs for the new ESXI hosts. In the example above, no corporate DNS server was specified, so the EVO:RAIL dnsmasq service is the source for all queries. It’s rare to actually need to modify this file, although one situation could happen is if you decide to change the management IP or FQDN of the servers listed here.
As for the forwarding of queries for systems not listed in the hosts file – that’s held in file dedicated to the dnsmasq configuration. So its not the usual /etc/resolv.conf file that usually holds the DNS Primary/Secondary IPs for Linux. The file used is called /etc/dnsmasq.conf held within the server= setting. We do have KB Article 2107249 (http://kb.vmware.com/kb/2107249) which describes the file and how to edit it. For instance you may wish to change the corporate DNS server entry if the IP address for the DNS service has changed or if you have fat-fingered the setting.
So to summarize. EVO:RAIL has its own DNS service that allows us to meet the requirements of vSphere for DNS. That’s ideal for greenfield deployments because we have no dependency on DNS, DHCP or Microsoft Active Directory. You can, of course point the EVO:RAIL DNS service to an ‘external’ or corporate DNS server for all other queries.
Actually, this happened last week – but I was so flattened by work leading up to VMworld – the joint announcement planned by myself and Carl fell flat on it face! That’s completely my mistake, as I totally dropped the ball on this one.
In case you don’t know the Ultimate Deployment Appliance (UDA) is a Community Project that I have promoted and used for some years – its an all-in-one PXE/DHCP/TFTP Appliance that massively simplifies the deployment of many operating systems – and I primarily use it for deploying VMware ESXi.
In my tests i found that merely selecting ESXi 5 Installable in the UDA menus and then selecting the ESXi 6 .iso worked right of the box. So it was a piece of cake for Carl to produce a patch bundle that allows you to select ESXi 6 from the menus to keep things both neat and logical.
The patch bundle can be download either from my site or Carl’s
I’ve been recently doing some scripting work with the Ultimate Deployment Appliance (UDA) which was developed by Carl Thijsen of the Netherlands. The reason for this work is to make it easy for me to switch between different versions of EVO:RAIL using my SuperMicro systems. I want to be able to easily flip between different builds, and its seemed like the easiest way to do this remotely was using my old faithful the UDA. This means I can run EVO:RAIL 1.2.1 which based on vSphere5.5, and then rebuild the physical systems around our newer builds, which incidentally use vSphere6.0.
Anyway, I encountered an odd error when scripting the install of VMware ESXi 5.5. One hadn’t seen with VMware ESXi 6.0. The error looked like said :Error: Error: Read-only file system during write on /dev/disks/naa.blah.blah.blah.
Would be enough to wipe any existing installation and VMFS volume. But the installer wasn’t happy. Incidentally “ST300MM0026” is the boot disk, a Seagate drive. However, that didn’t seem to work. I had to modify the ‘clearpart’ line like so:
I think what was happening was that clearpart wasn’t seeing the drive properly, and specifing it by model number allowed the VMFS partition to properly cleared.
Anyway, I doubt this will matter to most people, but I thought I would share in case someone else sees this…
UPDATE: Well, after automating the install of VMware ESXi 5.5, decided to flip back to VMware ESXi 6.0. I encountered the exact same error. So now both my 5.5 and 6.0 scripts include the change to clearpart.
EVO:RAIL is the first 100% VMware powered Hyper-Converged Infrastructure Appliance (HCIA). EVO:RAIL delivers compute, network, storage and management resources integrated onto an optimized 2U/4N hardware platform; all available via our 8 Qualified EVO:RAIL Partners and backed by single point of contact for both hardware and software. EVO:RAIL has gained a lot of momentum in a very short timeframe and the EVO:RAIL team brings continuously new capabilities to improve performance, scale, automation.
Join this session to get an overview of EVO:RAIL, a dive deep into the new EVO:RAIL 1.2 and a product demonstration from EVO:RAIL Product Marketing Manager and Product Manager. Presented by: Michael Gandy and Justin Smith, VMware
Around the end of 2014 the EVO:RAIL team released an update to their core software in the shape of the 1.0.1 release. One of the key features the release introduced was something we call “Link and Launch”, an optional feature used by our partners. As you might know, from a hardware perspective most EVO:RAIL appliances present pretty much the same amount of CPU/Memory/Disk and Network throughput – and that’s all to set to change with the announcement of more “flexible configs”. Some of our “Qualified EVO:RAIL Partner” (QEPs) differentiate themselves in the market place with their various software add-ons. EVO:RAIL’s “Link and Launch” feature gives our QEPs an engine to both automate the deployment of these add-ons, which often take the form of virtual appliances, as well offering links to these appliances. Sometimes these virtual appliances merely extend the functionality of the vSphere Web Client, at other times they offer a dedicated UI for managing the add-on.
The process begins at the factory. As you might know from reading this series of blog posts, node01 acts as a “bootstrap”, for want of a better word, for getting the EVO:RAIL appliance up and running. On node01 you will find the VMware “System VMs” in the shape of the vCenter Server Appliance and vRealize Log Insight Appliance. If the QEP is adding value with additional appliances they will be listed alongside the VMware “System VMs” and we often refer to these as QEP “System VMs”. In the screen grab below you can see vCSA, Log Insight alongside two ‘sample’ QEP VMs that I use to test this feature called “Test VM Number 1” and “Test VM Number 2”. These VMs would normally contain a product name and reference to the vendor. Notice also how neither the Log Insight nor these QEP VMs are powered on. They are only powered if needed (this is the case with Log Insight) or when the configuration of the EVO:RAIL completes (this is the case with QEP System VMs). We often refer to QEP System VMs that come with two components as the “Primary” and “Secondary” VMs.
Along side the QEP VMs we also get our partners to configure a small “manifest” file. This manifest file is a text file which contains friendly labels for populating the UI together with references to company logos such as the Dell or EMC logo. It’s this “manifest” file that populates the “QEP” section of the EVO:RAIL Configuration UI. In my case I used the generic “ACME” as the name of the vendor and QEP. In a production environment you would be more likely to see the vendor’s name such as HDS (Hitachi Data Services) or SMC (Supermicro).
Since the 1.0.1 release, when “Link and Launch” was made available to our partners, we have supported a new attribute to the JSON file. As you might remember from my other posts on EVO:RAIL it’s possible to have all the settings required for the EVO:RAIL Configuration engine stored in a text file with a JSON extension. EVO:RAIL supports the configuration of a single QEP System VM or two System VMs. In the screen grab below you can see the JSON file that I use in the hands-on-lab. If you look to the bottom you can see two additional, optional entries under the catagory of “vendor”.
It starts with the “vendor” attribute, and can be used to configure the two QEP VMs that have been imported into the system. Remember this is all done at the factory, so as a customer you merely need to provide your preferred IP for the QEP System VMs – and the EVO:RAIL engine will take care of deploying them for you.
Once the EVO:RAIL Configuration engine has deployed the appliance, at the very end it powers on the QEP System VMs and applies the IP configuration supplied. Once you login to the EVO:RAIL Management UI, you should see a “QEP” node in the left-hand sidebar.
In my case I just used a generic “ACME” style logo, and when you click to launch “ACME Test VM No.1” it just connects to a web-service.
This isn’t yet available to demo in our hands-on-lab, although I’m toying with the idea of including it in this year’s VMworld Labs. Our partners have already made great use of “Link and Launch” not least EMC, who have produced their own VPEX Blue management UI which has the look and feel of the core VMware EVO:RAIL Management UI.