Duke
thanks, it'll have to wait until we get the add'l power installed? I knew I was on the right track with what I was seeing.I suspected that mightbe the case.
I don't have time to read 702 page docs that are wrong in many places.I've seen enough of the HMC to know that it has poor wording on many of its pages.
installing vios via system template ,for instance, says 'click start when ready'.only they have no start button ( so I instead click on the 'next' button)..but how can you pass along a product like that? there was no QA of the HMC product? hmc 8.2 mind you.latest and *cough* greatest..
then they give you a read-only Vterm which stops at 'insert disk #2' but gives you no way to actually tell it to continue, because it is read-only.it also says click 'accept license' only you can't click the radio button. they have another section in documentation which clearly states that HMC will automatically find VIOS disk #2 in the hmc repository.it doesnt.it hasn't.it wont.LOL
outside of my shortcomings on the subject, it's painfully obvious that this is still a work in progress for IBM..the fact that you can't even 'vi ' a file in restricted shell is comical to me..
but as you say , I need the disk arrays powered up so it can see all of the disks and controllers, etc..
question: shouldn't this whole set up work on just the 8202 alone? I mean,what if I'm a client who doesn't need the extra 6+ TB disk space, I only need the 6x146GB internal drives on the 8202? I should then be able to install VIOS on it, and let it work with the internal disk controllers and network controllers and spread them out to the soon-to-be newly created VMs..right? so, in my case, why is this not working as such? I don't have the enclosures hooked up.the 8202 should work without a problem..
is there something that's telling the 8202 that it absolutely,positively needs the add'l enclosures to work? what if I had all three working, and decided to get rid of the g30 and the 24? I should still be able to work with PowerVM on the 8202 correct?
I distinctly remember when ordering the server that the enclosures were an add-on, not mandatory.so why can't I work without the add'l 210v power line, and without g30 and 24?
thanks for clarifying things though.you're right, no need to waste time playing with it.
If I sound completely off the rails frustrated , it's because I am..
dukessd is absolutely correct! You are doing it all wrong because you do not understand:
OK, but most critical: where are the (physical) SCSI adapters? The way you describe it you do not even have one and how are the disks supposed to be connected to the system, hm? dukessd said already that you probably have only one SCSI adapter and this is already taken by the first system you install.
No, i meant what i said: Linux virtualisation is: take some hardware, install Linux, install some software (for instance "Openbox") on top, then create several instances of running Openbox-processes (aka "virtual machines"), then install some OS of your choice into each of these virtual machines.
IBM virtualisation is: connect the MS to the HMC, create some LPAR profiles and install into each LPAR an OS. The "HW-host-OS-virtualisation-software" part of a virtualised Linux system is all in hardware.
Yes, because you seemingly have only one SCSI adapter, for chrissakes! You can add this adapter to several LPARs in the same way you can promise the same 10-Dollar-bill to several people. But finally, either you are going to have as many bills as people you made a promise to or all (except one) are going to be very disappointed. What you see is a LPAR which you have promised a SCSI adapter which it cannot get because you gave it to another system first. Switch off the first LPAR and you perhaps can boot the second LPAR without a problem (but as long as it runs you will not be able to boot the first - for the same reasons).
Well, i run some 40 managed systems with about 300 LPARs distributed on them, more or less the whole midrange IT of a big bank. I dare say if the system would be so bad as you describe that should not be possible. I might start with
which is a non-supported setup. You might not know that, because reading the documentation is below real experts, but this - along with several other necessary things you may not have done and things you didn't do according to the requirements - may also have led to your frustration. It takes very little creativeness from my side to come up with a possible course of action leading to the problems you encounter.
So you might want to reconsider your take on this and start with learning and only critisize things you have a well-founded knowledge about.
just to give you guys a status report, I've installed VIOS, and logged in to it with padmin..
I get what's going on now..it's basically the same as ldom creation in Oracle solaris.
ldm add-vdsdev
ldm add-vdisk
ldm add-vnet
similar concepts: backend file or volume.create the virtual device for it, assign it to a
LDOM ( or Lpar in IBM verbage), etc..
so it all makes sense to me now..
I can't see any virtual SCSI server adapters yet with lsdev -virtual, only the serial connections.I'll have to add them manually.I've created a VG with mkvg , a logical volume, etc..getting the hang of it.
we do have to get the power installed for the enclosures but at least I can mess around with this just to get my feet wet..
definitely prefer command line to gui.
yes, you have to add them manually in LPAR's profile definition. After that you have to _shutdown_ and start the LPAR again. Notice - not restart, but shutdown and start. If you have configured everything correctly and have luck, you can use Dynamic Partitioning -> Virtual Adapters menu in HMC without shutdown/starting the LPAR, but I don't believe. Anyway you should update your LPAR profile.
After that if you managed to everything right, you will see something like this in the output:
Hi everyone
KGB I have managed to create 2 AIX 6.1 VMs, as well as a VIOs server ( created it first). each has a rootvg of about 130GB in size , basically taking up one physical 136GB disk on the 8202..
.I have two virtual ethernet adapters going from VIOS to each VM .vm #1 works perfectly well, is networked and we're having no problems with it. AIX 6.1 VM #2 is unable to see the network.I have set up the DNS server with the hostname and the ipaddress ( 10.10.1.68,devrs8).
I have created add'l ethernet adapters on VIOS, rebooted it so they are 'seen' by LPAR #2, rebooted LPAR #2, and I've tried to configure ( on the AIX 6.1 OS) ,one and all at different times without any luck.they are there, they just don't like to ping any remote system.I can ping the above address I give it, or 127.0.0.1 but nothing else. def route has been set.
tried using ifconfig, smit, mktcpip, I even tried to use DHCP.no luck.even re-installed the OS to get a clean slate to work with, and I'm stuck at the same thing.
dns
10.10.1.6 10.10.1.7
IP=10.10.1.68
def route 10.10.1.1
netmask 255.0.0.0
The way you deescribe it the problem is not the adapters in itself. In typical setups you give the physical adapters to the VIOS(es), create virtualised adapters for the other LPARs and have the VIOS act as a virtual switch to connect these layers.
You being unable to get outside the system means you probably have got the setup of this part wrong or not at all. You need to set up VLANs both for internal and external networks and "connect" these on the VIOS. See the documentation, especially "SEA" and "HEA", for details about the concepts involved.
hi again
I think I've found the problem,on VIOS I have 2 virtual ethernet adapters, one set for 'bridged' connections the other not set in this manner.hmc 8.2
vm1 I has two virtual NICs one didn't work ( I'm assuming the one not set to 'bridged' on VIOS profile)
I will re-create one add'l ethernet adapter on Vios, select if for bridged access, and config LPAR #2 to use it..
have to say, I've read a lot of documentation and googled quite a bit about powervm networking.
it's a jungle.it really is confusing.
trying to relate this to the vmware world - in esxi,you go to configuration,and select physical vmnic, and add it to a vswitch(0) by creating a virtual network.it's actually quite easy.same with oracle solaris.the problem ariese with IBM and their convoluted use of the word 'logical'..
in theory, and correct me if I'm wrong, I should be able to use the two existing virtual NICs created on VIOS, the same ones that LPAR1 uses, and use them on LPAR #2,as long as LPAR 2 has two ethernet adapters installed in its profile. so vios doesn't need any add'l ethernet adapters created for a second LPAR. If I have 50 LPARS I need 50 virtual ethernet adapters created on the VIOS server's profile? no way..
it should be the same as putting a virtual machine on vmNetwork1 on VmWare esxi.you add a virtual NIC ( vmxnet3) , assing it to a virtual network.boot the OS, and configure it with DHCP or static IP.thats it.you can have many VMs using the same VNIC and it's related virtual network.same with oracle LDOMS.
I'm trying to put it in the vmware vocabulary so I can better understand it. IBM PowerVM Design for a VCDX | vcdx133.com
I'm apparently not the only lost soul out when it comes to PowerVM.
---------- Post updated at 01:22 PM ---------- Previous update was at 10:34 AM ----------
I figured it out,got LPAR #2 on the network, and can see every other server from it.
I see under the LPARs profile, the LHEA lists 4 etities. leave it to IBM to NOT call them NICs or give them names of any sort, like 'broadcom xxxx' , they just leave them as
U7aa.001.wzsg7kk-p1-c3-t1
" " t2
" "t3
" " t4
t1 and t3 were used for and on LPAR 1, but apparently VLANs were set to 'deny all' for this LPAR.
so I just took another NIC, t2, and allowed 'all' through it.
You might want to read this for help on configuring LHEAs. IMHO you should go for SEAs instead because these are more high-availability.
Anyhow, the "ugly names" you belittle are location codes. You can deduct from the location code exactly where a (physical) adapter has been placed and of which type it is. This information is much more valuable when you have a big system with several CECs and dozens of adapters and you want to replace one of these than the "Broadcom blabla" would be. It is really not IBMs fault that you seemingly have never had to administrate anything even midrange-sized, let alone the bigger irons.
You might have a very small system but it is built (and works) the same way as the dozen or so p780s that populate our data center. To work with a system that houses 50 LPARs with ease you need other methods than when you work with some PC. ESX basically run on PC hardware, somewhat rugged ip to be data center-ready. AIX does not. AIX runs on systems far bigger in every respect than any PC. Just have a look at the p780 and p870 datasheets to find out just how big these systems can get.
You might find the methods employed overly complicated but i can tell you from years-long experience that they are perfectly reasonable in an enterprise-level context. I suggest you either start using these methods and exploit the advantages they offer or you return to administrating your ESXi- or whatever virtualisation-platform you prefer.
As bakunin said, SEA - Shared Ethernet Adapters - are the way to go in an n+1 VIO environment for your client LPAR network availability.
Of course each VIO should also have n+1, like etherchannel.
Dual VIO allows for rolling, concurrent, upgrades to the VIO servers...
This is obviously a big step for you in the world of IT virtualisation, you are moving into a very different class so don't try and dumb it down to terms and other technologies you are familiar with.
You need to do the homework if you are going to get anything like the best out of this experience and the same applies for your companies investment in this kit.
