VMware Communities : All Content - All Communities

ESX deployment appliance v0.90

brugh — Thu, 12 Nov 2009 01:04:07 GMT

the whole point of the eda is that you don't need to build a general script.. it is specific for every server.

the IP address of the server is a stored in a variable called MyIP. if you need a second number you could change the 3rd octet like so:

secondIP=${MyIP/.2./.3.}

and then use $secondIP for the second cos port. so if your main ip address is 192.168.2.1, the secondIP would say 192.168.3.1.

another option is to not to put a second console port in but make the link redundant... better option than the extra management for a second cos port i think.

Hasslefree Download Appliance (HDA)

phntM — Fri, 19 Jun 2009 23:26:47 GMT

hi!

1st: i have been searching for an appliance like this (torrentflux)!

2nd: nearly everything works but: SHARE-space doesnt... :<

hostOS: winxp+sp3

share: c:\test - share name: test

host-ip: 192.168.1.100

created a new winxp-user: torrentlux with pass: torrentflux and gave maximum permissions and share permissions on: c:\test (share name: test)

firewall allows file-transfer

working: start => run => explorer
192.168.1.100 => prompt: user / pass - works...

what the hell do i have to enter EXACTLY in the config gui?!

Remote share configuration
Server => 192.168.1.100 // or =>
192.168.1.100 // or => smb://192.168.1.100 // or => ...
Share => \test // or => .\test // or => test\ // or => \test\ // or => ...
User => HOSTNAME\torrentlux // or => torrentflux // or => torrentlux@HOSTNAME
Password => this one is no prob (i think...)

thank you for your help in advance!

Why is RVI not ON by default?

AWo — Wed, 18 Mar 2009 20:35:18 GMT

Thank you. You answered all my question. Unfortunately I can't award points as I'm not the initiator of this thread...

AWo
VCP / vEXPERT 2009

ESX deployment appliance

marklemon — Tue, 17 Mar 2009 11:32:49 GMT

Strangeness here. I have the EDA all setup and working on servers in two of my computer rooms. Now I'm trying the first server in my 3rd computer room and although it's PXEbooting fine and showing the correct menu with my list of servers (so the nic in my server is working fine for PXEboot and working fine to get a DHCP address and even loads the boot menu from EDA).

My network admin confirmed that the port on the switch is configured correctly (the same as the other two that I just built with EDA beautifully!)

But then after "determining hostname" or something similar, it pops out the the manual config screen as it does when it doesn't like the kickstart configuration file.

"Choose a language". If I fill these all in and choose the exact same settings to get the install (NFS server ip address and NFS server directory path) it fails

I double checked the kickstart config file and it really hasn't changed, but when I look at the other virtual consoles on my new esx host I'm seeing:

Alt-F3:

doing kickstart... setting it up

waiting for link...

15 seconds.

reverse name lookup failed

ks location: http://172.x.x.x/ks/ks.php?hostname=zyxw

transfering http://172.x.x.x/./ks/ks.php?hostname=zyxw to a fd

failed to retrieve http://172.x.x.x//ks/ks.php?hostname=zyxw

trying to mount device hda

then

ALT= F4

loader: failed to set default route: Network is unreachable

My EDA is on the same subnet as the DHCP address that it picks up, as well as the ultimate static address that it'll be given, so I agree with my network administrator when he says the failed to set default route error doesn't matter and is a red herring.

I'm about to try it all over again, but with another nic in this server. In the mean time... any ideas anyone?

cheers,

Kevin

______________________________

Ah, when I did the install from CD I noticed that the list of nics are ordered oddly:

8:0:0 –e1000 82571EB Gigait Ethernet Controller

8:0:1 –e1000 82571EB Gigait Ethernet Controller

4:0:0 – bnx2 BroadcomNetXtreme II BCM5708 1000Base-T

42:0:0 – bnx2 Broadcom NetXtreme II BCM5708 1000Base-T

Maybe this is the wierd ESX reordering problem that I've read about. It would explain why PXEboot is getting network fine to load the EDA menu, but then the ESX installer is looking for eth0 and vmware has maybe ordered the nics so nic0 is one of my PCI nics which isn't configured for the service console. I'll try again using eth3 then eth4. Hopefully one of these will work.

Message was edited by: Kevin

unattended esxi

brugh2 — Tue, 09 Sep 2008 09:46:18 GMT

they're a great help in understanding how the esxi installation process works

i got a setup where it boots through PXE, installs (manually) and after a reboot i have a bare esxi server. so far so good, but i'm still missing something in the process as to where to exactly start customizing. i pass my custom.tgz and see it unpacked correctly in the busybox environment. but i'm unsure how to procede:

when the host boots, i press alt-f1 and login. i see the e1000 module loaded (have 2 nics in the host) but i dont seem to have network connectivity. is it supposed to? how do i configure it?
i guess i could rewrite the /usr/lib/vmware/somethingWelcome.py to unload the lpf and qla drivers, then find a harddisk and dump the visor image on it. seems a bit eerie though. are there hooks or other scripting parts that allow for a supported way to pass (unattended) configuration parameters to the installer?
i added some parameter settings after vmkernel.gz on the append line. but i'm not sure how to use those in the script i put in /etc/rc.local.d. what's the secret here?
customization of the host can't start until the dd image is put on the harddisk. again this would require a rewrite of the installer .py scripts which i would rather not do (mainly for compatibility reasons). is there a supported way to write to the newly written esxi host disk?
i see you use a midwive to automatically configure a new esxi server. but as far as i can see, it starts configuring the busybox environment. this would not change the esxi host. but i'm sure it changes the host, not the installer. but what am i missing here? this would probably be a clue to how my setup would have to work also.

any help would be greatly appreciated!

ESX 3i update 2 missing 'shell' ??

Dave.Mishchenko — Tue, 02 Sep 2008 02:25:36 GMT

The shell option was removed quite a while back (I think after the beta). The option mentioned above is what works now and you can enable SSH as well (unsupported though) http://www.vm-help.com/esx/esx3i/ESXi_enable_SSH.php

ESX deployment appliance v0.80

brugh — Tue, 26 Aug 2008 19:44:22 GMT

i updated the EDA to v0.80. check it out on http://virtualappliances.eu/.

please post any comments on http://communities.vmware.com/thread/139465?tstart=0&start=75

Ubuntu 7.10 Just Enough OS appliance

Lance Rushing — Wed, 06 Aug 2008 19:11:09 GMT

For Ubuntu Hardy 8.04 I got eth0 working by doing the following:

sudo rm /etc/udev/rules.d/70-persistent-net.rules
sudo reboot

Also I went back and removed the 70-persistent-net.rules from my gold source used for cloning.

I suppose you could make a script that would remove 70-persistent-net.rules on each boot/halt. Then you can roll out the appliance and not worry about the eth0.

Ubuntu 8.04 bootstrapped JeOS

brugh — Fri, 18 Jul 2008 15:35:49 GMT

Yes, it's jet another JEOS appliance based on Ubuntu. Only this one isn't just a default install of the jeos.iso available at ubuntu.com but a new appliance, built from scratch with deboostrap. After adding a virtual kernel and vmware tools it got cleaned up and the harddisk rebuilt, leaving a perfect base to built any new virtual appliance on.

Check out http://www.vmware.com/appliances/directory/1335.

Fast and easy ESX installer!

brugh — Sun, 27 Apr 2008 09:48:12 GMT

it's totally GPL licensed so it's free. but if you want to make a donation for all the time it will save you, let me know

and i didn't post this as a question, so i can't award points. sorry

what makes an appliance just a VM and when is it a true appliance

brugh — Tue, 26 Feb 2008 17:07:16 GMT

there's something that's been bugging me for a while. i'm curious if i'm the only one so here's the thing:

i see a lot of people submitting appliances that in my book do not account as one.

appliances are supposed to be mean and lean and be stripped from all superfluous bagage. if somebody takes a CD, installes it in a VM and submits it, it's by far an appliance. it's just a VM. it has a very large attack surface, needs lots of patching and doesn't add anything. anybody can do a default installation in a VM.

to be an appliance i say you'd have to do at least the following:

install the OS
install VMware tools
install the application(s) it's supposed to run
optimize / strip / clean the installation
clean caches
clean pagefiles / swaps
add new harddisk, copy all files to that one, make that the primary disk (decreases it's size by 2 or 3 times if it's a growing disk. compressing with vmware's tools doesn't work)
stop the VM and strip the .vmx of all IDs and mac addresses

once that's done, and no sooner, would i give it an 'appliance' stamp.

the second to last step is probably somewhat tricky for a lot of people. here's a quick howto of what you would do on a linux box:

add a new harddisk
partition it according to the first disk's setup and create the filesystem/swapspace
mount it under /mnt or something
copy the files as bare as you can with a command like: find / -mount -print | cpio -pm /mnt
now edit /mnt/boot/grub/devices.lst and change the 'sda' to 'sdb'
run 'grubinstall --root-directory /mnt'
now edit /mnt/boot/grub/devices.lst and change the 'sdb' back to 'sda'
halt the machine, remove disk 1 completely, make disk 2 the new disk one (change scsi0:1 in the vmx to scsi0:0)
now zip the vmx and vmdk files and you have a decent appliance.

i hope this helps people to make more sensible appliances.

i'd appreciate a reply from anybody who agrees.. or disagrees.. i just hope it ups the value of the general appliance in the directory.

New: download appliance (torrents + usenet)

brugh — Wed, 09 Jan 2008 00:40:48 GMT

I have a new download appliance that can download torrents and usenet binaries, automatically par2 repair them, unpack them and put them on a windows share (mediacenter ;)).

I'd like to share it with everybody but I have no means to put it on internet 24x7 (my webspace is full). Does anybody have a spare 250MB to host it on? Please drop me a message.

new appliances

brugh — Sun, 06 Jan 2008 22:53:06 GMT

Never mind, i found it: http://www.vmware.com/appliances/directory/create_community_appliance. kinda obscured but it was there.

Out now: X-monowall 1.232

brugh — Sat, 05 Jan 2008 11:54:28 GMT

check out http://www.vmware.com/appliances/directory/150

changes since the previous version:

captive portal reliability fixes
- ensure that the pruning process is always run on all active users
- properly handle sessions that have not passed any traffic by the time they end
- improve locking
fixed FIN handling in ipnat FTP proxy
updated timezone data

and it also has a newer version of the guestdaemon which in previous releases crashed under suspicious circumstances so it should be even more stable now.

i also have a version running on the 1.3 beta release so as soon as that's final on http://m0n0.ch i'll have a x-mono version soon after that. should be good .

thanks for everybody's support. have fun!

oh, and please post any questions regarding this appliance in the http://communities.vmware.com/thread/43915?tstart=0&start=100 thread

curious: VirtualCenter as a Virtual Machine

brugh — Mon, 06 Aug 2007 21:28:35 GMT

ok, seems the problem's been solved. seems there were routing issues to and from the esx servers to VI clients. stupid network guys once that was fixed there were no more serious issues. still odd it can go wrong like that

ESX 3 is missing wget

jasonboche — Fri, 15 Jun 2007 14:54:29 GMT

You can use lwp-download, is provided by esx3 and
work perfectly with big files.

lwp-download works great. I've been using it for a few months in order to script my patch installations from a web repository. Just remember you have to open up the outgoing firewall in order to use it

esxcfg-firewall -allowoutgoing

ESX 3 forces vlance?

oreeh — Mon, 16 Apr 2007 15:15:21 GMT

With VC you have to unregister (yes that's remove from inventory).

\ponder can license services for Citrix and VMware be combined?

hgoldman — Mon, 06 Nov 2006 21:21:43 GMT

It can sometimes be a problem combining them since it can clutter up the logs.
From a technical point of view it would be better to run them side by side on the same machine using each their tcp-port.

You can use our software License Statistics then to both monitor when the servers are up and down (e.g. get emails when servers go down) and track the actual license usage. http://www.x-formation.com/license_statistics/index.html

-- Henrik

To Hyperthread or Not to Hyperthread

devzero — Wed, 01 Nov 2006 00:24:53 GMT

i think you may get better help on a linux related mailinglist. first try the distro-specific lists/forums - if nothing helps, you will be helped on LKML (Linux Kernel Mailing List)

How much is true of the 50Mbit SC limit

brugh — Fri, 27 Oct 2006 14:38:00 GMT

nobody ever seen this!?

minimalistic linux kernel

king@it.ibm.com — Wed, 23 Aug 2006 07:36:40 GMT

Well the hypervisor pretty much exposes a standard (yet virtual) hardware interface so streamlining a Linux kernel to it wouldn't be much different than streamlining a Linux kernel for all other hardware (physical or virtual).

Your post and specifically "doesn't need to do all sorts of fancy stuff that the hypervisor will take care off" implies that you are aware of the fact that many low level functions are performed already by the hypervisor and those are also performed at the Guest level (since the guest doesn't know that it's running on an active/intelligent piece of software). The solution to that is called paravirtualization (i.e. making the vmkernel aware of the fact that it's running on an hypervisor and not on a real piece of hardware) but today VMware does not support this (they said they will support it down the road).

Massimo.

web- and email filter

brugh — Mon, 14 Aug 2006 21:34:49 GMT

glad i didn't go through the trouble i'll keep it to myself

why do idle VMs eat CPU-time

brugh — Fri, 07 Jul 2006 21:35:54 GMT

no it's a linux ubuntu5 32bit installation

build 23869 breaks vmware-authd

ShaneV — Thu, 06 Jul 2006 14:58:59 GMT

I had this same problem and found that that when I built the latest library it went away.

Are ESX3 virtual switches really managed?

Deli — Fri, 16 Jun 2006 09:22:11 GMT

Hi brugh,

what do you want to manage?

I had not missed anything in 2.5 or 3.0.

Regards,
Marco

Debian hosts have an old libssl

brugh — Mon, 05 Jun 2006 18:24:47 GMT

i have inetutils-inetd running. updated to the xinetd package and voila, works again. thanks for the advice, put me on the right track.
but it's weird that beta2 worked and updating to beta3 broke it. as far as i know the host hasn't changed in that time, xept for the usual OS updates. but it works again, thanks.

Add VMXNet and Heartbeat to Monowall!

martinwa — Mon, 05 Jun 2006 16:45:22 GMT

Hi brugh (and others),

I am using m0n0wall from your web-page, it works great.

But does anyone has a working copy witch has hart-beat running?

Regards
Martin

Error writing to vmware-authd socket. error 10013

brugh — Mon, 15 May 2006 14:59:52 GMT

nice, this one goes into my personal knowledgebase

my solution was simpler though.. i just reinstalled the client and after that, it worked again :P who'd have guessed

ESX multipathing failback timeout

MR-T — Fri, 12 May 2006 11:46:09 GMT

I would assume it uses the same metrics to check when the paths are fixed. (This is only my assumption).

If it's scanning to see when the connections aren't there, it's fair to assume it can kill 2 birds with 1 stone and see if they've returned.

Add VMXNet driver and heartbeat to M0n0wall

brugh — Mon, 13 Mar 2006 21:08:26 GMT

it works.
not only vmxnet drivers work but i got the heartbeat to work also. i think it's pretty cool

Strange CPU behaviour

brugh — Wed, 08 Mar 2006 22:46:12 GMT

gsx performance problems ..