+ Reply to Thread
Page 2 of 4 First 12 34 Last
Results 26 to 50 of 83
  1. Senior Member
    Join Date
    Apr 2012
    Location
    Montreal area, Qc Canada
    Posts
    156

    Certifications
    VCAP-DCA/DCD;VCP5-DCV;MCSE 2012;CCNP;CCNA Security;Security+;A+
    #26
    Seems like you are using vApps to "organize" your structure, is that the case or you actually need them to start in sequence?
    Reply With Quote Quote  

  2. SS -->
  3. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #27
    Thank you very much for taking the time to check it over and questioning. I hope more is to come

    A lot of people use vApps or even Resource Pools to organize the structure, as folders are nice but won't appear on the cluster / host view (why not ey).

    But to answer the question, I only use vApps to indeed organize the startup order. On single hosts you can obviously configure automatic startup and power-off, but in clusters you can't.

    As I am working in a hosting company I don't REALLY need to power-off any servers. But at home, as the whole lot uses approx. 6A, I do, and mostly power it on / off using remote controller power bars.

    So out of habit I configure vApps solely for start-up order. So I need to make sure that certain servers booting up in a particular order.

    1. DCs
    2. SQL
    3. VCS
    4. Virtual hosts (if applicable)
    5. VSM (if applicable)
    6. VCC (if applicable)
    7. The rest

    So as a result the vApps are configured as follows (hope it makes sense, so imagine when you pulled the plug and turn it back on)

    Core Infrastructure - needs to come up in a certain order



    Random stuff which don't need to come up in any order



    Take the vCloud Infrastructure - virtual hosts first, then the rest



    For the hosts, I don't care which order



    And for the actual vCloud infrastructure - based on importance

    1. vCenter
    2. vShield Manager
    3. vCloud Cells



    Like I say - I really appreciate the input and correction if / when needed
    Reply With Quote Quote  

  4. Senior Member
    Join Date
    Apr 2012
    Location
    Montreal area, Qc Canada
    Posts
    156

    Certifications
    VCAP-DCA/DCD;VCP5-DCV;MCSE 2012;CCNP;CCNA Security;Security+;A+
    #28
    Makes sense for the DC, DNS, VCenter etc... However... for the rest, where start order isn't important, you just introduced a major issue in your infrastructure. You probably should create a single Resource Pool and put all those "any order" VMs in it and create a powershell script to start them in the order you want. (Really looking forward for VMware to offer start order across cluster instead of just host).

    Using vApps or resource pools for any kind of "organising" is a bad idea. vApps are resource pools, and are viewed as an entity in itself. In case of resource contention, the shares allocated to either pools or vApps will define how much resources the VMs inside them will get to use.

    You have to make sure you balance the shares allocated to the vApps, Resource Pools and VMs appropriately.

    For example, take a vApp with multiple VMs inside it, and a single VM with 4 CPUs at the same level. Both get 50% of the resources of the host (4000 shares each). If you vApp had 10 VMs in it, they would each get 400 shares vs the 4000 for the VM at the same level as the vApp.

    Resources.jpg
    Reply With Quote Quote  

  5. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #29
    I must admit that performance is my weak point so I appreciate your input - thank you very much ..

    So basically I am "alright" until I have additional VMs. Because right now, taking the "Production" as example

    - vApp LN1
    -- vApp ESX Hosts
    -- vApp Management

    That would translate to

    - vApp LN1
    -- vApp ESX Hosts - 4000 Shares
    --- c1b01 - 2000 Shares
    --- c1b02 - 2000 Shares
    --- c1b03 - 2000 Shares
    -- vApp Management - 4000 Shares
    --- VCS --- 2000 Shares (40%)
    --- VSM --- 1000 Shares (20%)
    --- VCC01 --- 1000 Shares (20%)
    --- VCC02 --- 1000 Shares (20%)

    So If I were to add a big VM into the "root", I'd be messing up the shares in the vApps causing VMs inside the vApp to receive less shares maybe causing performance issues ....

    But I really wished as well that we could have startup orders without the use of vApps in clustered environments
    Reply With Quote Quote  

  6. Senior Member
    Join Date
    Apr 2012
    Location
    Montreal area, Qc Canada
    Posts
    156

    Certifications
    VCAP-DCA/DCD;VCP5-DCV;MCSE 2012;CCNP;CCNA Security;Security+;A+
    #30
    Quote Originally Posted by jibbajabba View Post
    I must admit that performance is my weak point so I appreciate your input - thank you very much ..

    So basically I am "alright" until I have additional VMs. Because right now, taking the "Production" as example

    - vApp LN1
    -- vApp ESX Hosts
    -- vApp Management

    That would translate to

    - vApp LN1
    -- vApp ESX Hosts - 4000 Shares
    --- c1b01 - 2000 Shares
    --- c1b02 - 2000 Shares
    --- c1b03 - 2000 Shares
    -- vApp Management - 4000 Shares
    --- VCS --- 2000 Shares (40%)
    --- VSM --- 1000 Shares (20%)
    --- VCC01 --- 1000 Shares (20%)
    --- VCC02 --- 1000 Shares (20%)

    So If I were to add a big VM into the "root", I'd be messing up the shares in the vApps causing VMs inside the vApp to receive less shares maybe causing performance issues ....

    But I really wished as well that we could have startup orders without the use of vApps in clustered environments
    Well, until the resources are in contention, you're fine. However...

    - vApp LN1 (4000 shares)
    -- vApp ESX Hosts - 4000 Shares
    --- c1b01 - 2000 Shares
    --- c1b02 - 2000 Shares
    --- c1b03 - 2000 Shares
    -- vApp Management - 4000 Shares
    --- VCS --- 2000 Shares (40%)
    --- VSM --- 1000 Shares (20%)
    --- VCC01 --- 1000 Shares (20%)
    --- VCC02 --- 1000 Shares (20%)

    Your resources are split 3 ways, so the whole cluster is split with 33% of the resources for LN1, 33% of the resources for ESX hosts and 33% for Management.

    Out of that 33%, your hosts (3 of them) are getting 33% of what the vApp has, so basically each host would get 11% of the cluster resources. In your management vApp (which gets 33% of cluster resources), each VM would get it's share of what the vApp has (so VCS would get 40% of 1/3 of the cluster resources).

    You can leave them as vApps, but I would adjust the shares more appropriately. Try to figure this out as if you were running production on this

    Have fun!
    Reply With Quote Quote  

  7. kj0
    kj0 is offline
    Apple and VMware kj0's Avatar
    Join Date
    Apr 2012
    Location
    Brisbane, Australia.
    Posts
    744

    Certifications
    vExpert x 4 | Apple Mac OS X Associate | Cert III - IT.
    #31
    Eval Ran out, so I now have to rebuild my lab to so I can get some practice in before my VCP exam.

    Your lab layout has given me some great ideas on how I should set my new lab up.


    I'm liking all the extra bits you are throwing into this lab. Keep up the excellent work, JJ.
    2017 Goals: VCP6-DCV | VCIX
    Blog: http://readysetvirtual.wordpress.com
    Reply With Quote Quote  

  8. Self-Described Huguenot blargoe's Avatar
    Join Date
    Nov 2005
    Location
    NC
    Posts
    4,098

    Certifications
    VCAP5-DCA; VCP 3/4/5/6 (DCV); EMCSA:CLARiiON; Linux+; MCSE:M 2000/2003; MCSE:S 2000/2003; MCTS:Exch2007; Security+; A+; CCNA (expired)
    #32
    Quote Originally Posted by jibbajabba View Post
    But I really wished as well that we could have startup orders without the use of vApps in clustered environments
    I share the same gripe. It's not often that a full power down of VMs is necessary, but sometimes it is, and when you have 100's of VMs and need certain ones online before others, it would be helpful. Also, if you have a distributed application across several VMs that requires a reboot order, but you don't want to impose the "resource pool" effect on it.

    Anyway, have fun in your VCAP journey, you will learn a lot.
    IT guy since 12/00

    Recent: 10/27/2017 - Passed Microsoft 70-410 (one exam left for MCSA 2012)
    Working on: MCSA 2012 upgrade from 2003 (to heck with 2008!!), MCSA 2016 upgrade, more Linux
    Thinking about: VCP6-CMA, AWS Solution Architect (Associate), Python, VCAP6-DCD (for completing VCIX)
    Reply With Quote Quote  

  9. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #33
    Thanks guys. As always, this place is great for encouragement

    My progress is somewhat slower than expect at the moment. I need to upgrade out infrastructure, including vCloud Director, from 5.0 / 1.5 to 5.1 so my lab is somewhat bastardised at the moment as we don't have any testing kit at work (I know right). So testing the upgrade beforehand is obviously in my own interest. So my lab now has also vCloud Director 5.5 running alongside my nested 1.5.1 one .. Oh and for good measure I threw Operations Manager in as well (never used it, but I like it).

    Then I did some hardening of my hosts, installed a public / private key combo so you won't be able to login via SSH without the key. The servers are joined to the domain as well but what I haven't tested yet is whether you can create public / private key combos for domain users .. Should make no difference, but might test that too.

    I also ordered a SSL certificate so I can play "replace that self-signed thing"

    You know - my approach of studying for the VCAP is something I should have done with the VCP lol ...Seem to be doing a lot of stuff I should have done years ago .. but there you go...
    Reply With Quote Quote  

  10. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #34
    Massive set-back. I created a provider VDC on my vCloud 5.5 setup which threw every single host into maintenance mode. What a stupid thing is that? Needless to say it didn't go well. Saturated the switchports of course and everything came to a halt. Not to mention that every single VM tried to migrate to another host (guess how successful that was with every host going into maintenance mode) ...

    Only box still accessible is the storage server as it is physical.

    Had to pull the plug so will need to see if I can recover it later.

    Meh ... One disadvantage of using vDP for backups is you need a vCenter for restores to begin with and I am not sure I can be bothered. We'll see once I had a tea and some breakfast
    Last edited by jibbajabba; 02-20-2014 at 09:27 AM.
    Reply With Quote Quote  

  11. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #35
    Thankfully a complete powercycle of the whole infrastructure brought it back to life ... Man I hate moments like that ...
    Reply With Quote Quote  

  12. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #36
    Not going well at the moment - so busy at work it's unreal .. I am leaving the house at 6am and I am usually not back until 8pm ..

    Only productive thing I have done is keep reading "Mastering vSphere 5.5" and go through all Advanced videos on Pluralsight ..

    And registering a new domain : vcap.guru

    Ah well ... No need for a marathon to get to the finish line - crawling get's you there eventually too
    Reply With Quote Quote  

  13. Senior Member tprice5's Avatar
    Join Date
    May 2012
    Location
    Nashville
    Posts
    758

    Certifications
    BS:CS, VCP5-DCV, Sec/Svr+, MCSA:08/12, MCITP:SA/EA, CASP, VCA:WM/DCV/Cloud, ITIL-F, MOF, ISO/IEC 20000/27000
    #37
    Quote Originally Posted by kj0 View Post
    Eval Ran out, so I now have to rebuild my lab to so I can get some practice in before my VCP exam.
    Stanly provided you with a license key. Why didn't you just use that?
    Reply With Quote Quote  

  14. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #38

    Default iSCSI MPIO using VDS / 2 Uplinks / 4 hosts - somewhat confused now.

    Right, I am having a brainfart here and it seems I am not the only one

    So I thought I throw this out here for discussion (even though there is the risk of people telling me off for being stupid haha).

    ISCSI Portbinding ... Now back in the days you likely configured MPIO using standard switches

    VMK1
    - VSS1
    -- VMNIC1

    VMK2
    - VSS2
    -- VMNIC2

    Then you use the CLI to MPIO the lot.

    Right, now vSphere Distributed Switch.

    Portbinding is now fairly easy, right.

    So host has two uplinks. You create two portgroups



    and you make sure each portgroup is using one particular uplink



    Then of course you use these two VMKs for the software initiator



    Baddabing, baddabush, you are ready to go.

    So where's the confusion ? Simple, a friend of mine needed to check something so I lend him my lab .. ok .. he was all over it when he saw my setup and called me names (ungrateful bastid )

    So my setup :



    As you can see, I have two iSCSI / vMotion portgroups per host. Now he was telling me that this is nonsense. He isn't the only one though .. now a few people told me that you won't need two portgroups per host. You just do what you do with the two ISCSI portgroups, bind them to uplinks and done.

    That is fair - and I guess makes sense, but if you do that - all remaining hosts won't see their VMks as compliant as the relevant portgroups are in use by other VMKs / ISCSI initiators.

    I can see that it might be somewhat rubbish to have 50 hosts and you end up with 100 Portgroups, but ... I don't know what "but" - But until now I thought I knew what I was doing and now I am more confused then anything ..

    Any thawts ?
    Reply With Quote Quote  

  15. DoWork
    Join Date
    Jun 2010
    Location
    A major Illinois hospital system near you
    Posts
    1,468

    Certifications
    vExpert, VCAP5-DCA/DCD, VCP5-DCV, VCIX-NV, VCP-NV, BSTM
    #39
    You don't need two per host in a dvSwitch



    Remember, the VMKs are unique to the host in question because they are part of the dvUplinks group. So each one is completely independent of the other hosts. They will all show compliant.
    Last edited by QHalo; 02-24-2014 at 06:45 PM.
    Reply With Quote Quote  

  16. Learn it, Do it, Know it! Asif Dasl's Avatar
    Join Date
    Aug 2010
    Location
    Ireland
    Posts
    2,061

    Certifications
    MCSA:2008/7 MCITP:EA/SA EDST/EST MCSE'03 SBS'08/'03 MCDST A+ N+ Sec+ Srv+ VCP5-DCV VCA-DCV/Cloud
    #40
    I will admit I am still a noob, so with that said and full expectation that someone might correct me... Here's how I set mine up (Post #19 the middle picture) - I was under the impression that the vmkernels & IPs are different on each host for MPIO but you use the same vDS portgroup for iSCSI / management / etc. I just set mine up for all traffic because it's almost a 10GigE connection with 9 NICs but I know some people split it up like this...

    HTH
    Reply With Quote Quote  

  17. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #41
    @QHalo, so you aren't doing any port binding at all then? Because if you create a vmk port, assign it to a port group and bind it to an iSCSI initiator, it won't be available to any other host / iSCSI software initiator.
    Last edited by jibbajabba; 02-24-2014 at 06:49 PM.
    Reply With Quote Quote  

  18. DoWork
    Join Date
    Jun 2010
    Location
    A major Illinois hospital system near you
    Posts
    1,468

    Certifications
    vExpert, VCAP5-DCA/DCD, VCP5-DCV, VCIX-NV, VCP-NV, BSTM
    #42
    I'm absolutely doing port binding with all my iSCSI ports on all my hosts.
    Reply With Quote Quote  

  19. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #43
    Quote Originally Posted by QHalo View Post
    I'm absolutely doing port binding with all my iSCSI ports on all my hosts.
    Right, like I say, it all makes sense but something ain't right. I'll have another look tonight or tomorrow. Feels like I never worked with vSphere lol.
    Reply With Quote Quote  

  20. DoWork
    Join Date
    Jun 2010
    Location
    A major Illinois hospital system near you
    Posts
    1,468

    Certifications
    vExpert, VCAP5-DCA/DCD, VCP5-DCV, VCIX-NV, VCP-NV, BSTM
    #44
    Let me know if you want another set of eyes. We could do a webchat or something.
    Reply With Quote Quote  

  21. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #45
    Appreciated but I need to find that out myself.

    God, played too much with FC over the last few years ... losing the edge (and mind) here lol.
    Reply With Quote Quote  

  22. Senior Member tomtom1's Avatar
    Join Date
    Feb 2014
    Posts
    374

    Certifications
    JNCIP,SP, JNCIS-SP, CCNP, VCAP5-DCA, VCP5, MCITP 2008 SA, CCNA
    #46
    Having multiple vMotion VMkernel interfaces allows for loadbalancing over the pNICs. If you have multiple VMkernel ports dedicated to vMotion, you can create a 1 to 1 mapping of the VMkernel port and the pNIC. This achieves better loadbalancing (simultaneous data streams) then placing all NICs under the Active Adapter heading of the failover order.

    If you do this, I would suggest using 2 port groups for the vMotion network on the DVS. In my lab (which I used to achieve VCAP-DCA), I set it up like this:

    Screen Shot 2014-02-24 at 21.16.50.jpg Screen Shot 2014-02-24 at 21.17.25.jpg

    So, there is a method to the madness, but unless you specifically want this, you should be fine with just one vMotion enabled VMkernel port.

    See also: http://kb.vmware.com/selfservice/mic...rnalId=2007467
    Reply With Quote Quote  

  23. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #47
    Quote Originally Posted by QHalo View Post
    Let me know if you want another set of eyes. We could do a webchat or something.
    No idea what happened, but all working

    I think my first hunch was the best describing the situation:
    Quote Originally Posted by jibbajabba
    Right, I am having a brainfart here
    Reply With Quote Quote  

  24. DoWork
    Join Date
    Jun 2010
    Location
    A major Illinois hospital system near you
    Posts
    1,468

    Certifications
    vExpert, VCAP5-DCA/DCD, VCP5-DCV, VCIX-NV, VCP-NV, BSTM
    #48
    Cheers bub. Glad you worked it out.
    Reply With Quote Quote  

  25. VCDX in 2017 Essendon's Avatar
    Join Date
    Sep 2007
    Location
    Melbourne
    Posts
    4,489

    Certifications
    VCIX-NV, VCAP5-DCD/DTA/DCA, VCP-5/DT, MCSA: 2008, MCITP: EA, MCTS x5, ITIL v3, MCSA: M, MS in Telecom Engg
    #49
    Quote Originally Posted by kj0 View Post
    Eval Ran out, so I now have to rebuild my lab to so I can get some practice in before my VCP exam.

    Your lab layout has given me some great ideas on how I should set my new lab up.


    I'm liking all the extra bits you are throwing into this lab. Keep up the excellent work, JJ.
    Bit late to post this and kinda irrelevant to the thread, but you can "renew" the eval license (certainly not supported for production, labs only). Check this out > Information about Cloud Computing (IaaS) and VMware Technologies: vCenter Reinstallation after 60-Days trial expires without losing Database Contents
    VCDX: DCV - Round 2 rescheduled (by VMware) for December 2017.

    Blog >> http://virtual10.com
    Reply With Quote Quote  

  26. Google Ninja jibbajabba's Avatar
    Join Date
    Jun 2008
    Location
    Ninja Cave
    Posts
    4,240

    Certifications
    TechExam Certified Alien Abduction Professional
    #50
    If your company is VMware partner then there might be a chance they got NFR licences.
    Reply With Quote Quote  

+ Reply to Thread
Page 2 of 4 First 12 34 Last

Social Networking & Bookmarks