When root partition is full on ESX host

I had been in a strange situation when upgrade of ESX host was failing. It was giving me all kind of wired messages.
This host had NFS mount where I had placed my swap files. Host was suppose to moved from one location to other location. Person who powered off did not realize that it had NFS mount. When ESX host was powered on different location it started giving me error message that it is still trying to do NFS mount (ALT+F12 log).

I was also getting error message "Configuration changes were not saved successfully during previous shut down" . The reason for this was ,root file system was full (Please dont ask me why , we have many admin)

I used "esxcfg-nas " command to delete that mount. But still my root partation getting filled message didnot resolved.

So I thougth of running following command to find out what files are above 10GB.

find / -size +10240000c -exec du -h {} \; | less


I found that some one has stored ISO file under /vmimage . :)

Deleted thoso ISO and I was able to upgrade my host to 3.5U4

My new status in VMTM forum.

I had been member of VMTM forum since 2007. I had recently given status of “Hot Shot”. It really thrill me.

Level avilable inside VMTM's are

My design for ESX network

I had been trying to find out what would be the best way to configure NICS on ESX.

My college told me to use Tagged VLAN for each component. Then I got into trouble that I cannot use single switch to have two tagged VLAN associated and use that for additional redundancy.

I argued and told why you want to tag VLAN when it got trunked on my NICs instead you label it as VLAN1 (In layman term trunk all the VLAN which has been created) and if I wish I will tag the packets when I send it out of my vSwitch.

What I have read about tagging is “ It is just a help you in switch Administration and have nothing to do with technical advantage” . If that is the case then my dear network engineer you trunked every VLAN on each of the NIC and I will pick and choose. This will be very helpful for me to design.

If you want to achieve this then you need to have following config on each of your port associated on your vSwtich

description * ESX*

switchport access vlan 1

switchport trunk encapsulation dot1q

switchport mode trunk

spanning-tree portfast trunk

I have also shared my one of the network design

When Interviewer does not know their own stuff like Capacity Planner is Free?

I keep attending technical interview for various reason. I came across an interviewer who happens to be a working with big Indian MNC and who was interviewing me on VMware technologies. I don't claim myself as a "punter" but I assume I know my stuff to some extent.

He asked me if I have used capacity planner. To this I told yes I have used but in conjunction with VMWare TAM since these are paid service. He told CP is free. Well it is free but you can not use its functionality until you paid for this service. Which means you can download and install but won't be able to explore what it is suppose to do. I was absolutely correct

I wrote to cpprovisioning@vmware.com and this is what they have to say

"Yes, Capacity Planner is a free product but in order to get it you must attend the 2 day instructor led course which is not free. Retail cost is $1495. Please speak with a sales rep in order to see about a lesser price."

Please find the URL which explains the same

http://vmetc.com/2007/12/06/how-do-i-get-vmware-capacity-planner/

http://www.vmware.com/products/capacity_planner/

He then asked me what does the VMware recommend for lun size? I told VMware does not recommend any lun size. It's storage vendor who provide recommendation. And it make sense because that's vendor who represent lun, vmware just provide interface. This was discussed here http://communities.vmware.com/thread/104211

To this he started telling it is mention under best practice document from VMware . GOD why people want to prove themselves? This is just a interview and you are interviewing a person. It may happen that he knows stuff more than you but that does not mean that you will try to misdirect him. I really don't care for such person and neither would I like to work with such person.

A new mile stone:Automated ESX deployment using virtual appliance EDA/UDA

I had been struggling to find out way to automate my ESX build and found two appliance using which you can automate ESX deployment. Well this not only the way you can deploy but there are folks who have many way to deploy scripted ESX . Like ADS to deploy ESX but that’s not the goal of this document. I want to share my challenge while setting up this appliance. I know the folks who are Champions of these appliance like Carl Thijssen/Mike Laverick/ Brugh have explained many steps on their blogs and forum.

Environment preparation before setting up PXE appliance for ESX deployment:

1. Dedicated VLAN just for ESX deployment so that you won’t disturb any of your production environments.

2. When you create DHCP scope for this VLAN make sure that you have reserve IP address as you are going to use few IP for PXE /ESX/Appliance.

3. This VLAN will be Native VLAN.

4. One ESX host create a separate vSwitch and dedicate NIC for this appliance alone. Assuming that your appliance going to run out of ESX host. Set this NIC to 100MB half. The same setting you need to do on switch side of that port

              switchport trunk native vlan xxx

              switchport mode trunk

          speed 100 duplex half

          spanning-tree portfast trunk

After appliance has setup on ESX HOST:

5. After you install appliance follow the instruction on their respective site for configuring appliance. If you are not using DHCP from appliance then make sure that you have stopped it. If you have not stopped then also you land of on different issue. It is better to use windows DHCP. If you have setup your appliance on same VLAN as ESX host to be build then you don’t have to do any changes on DHCP server as someone has already explained. On ESX build VLAN just add IP Helper address for appliance. Yes this has worked for me.

6. Also do clearly specify DNS /Gateway IP while configuring appliance. If you miss any of these setting then also you may land up in problem while deploying appliance. I thought since it is testing so these are not required but I was wrong as I landed up in trouble latter.

7. There is also challenge to decide on Kick start (KS) nic. Normally your kickstart device should be your onboard NIC. Try some permutation and combination to decide on NIC. In my example if I had 10 NIC amongst which we have two onboard NIC then my onboard NIC device becomes eth8/eth9. Yeah it is quite surprising and will update when I found. But for now what I do is try permutation combination to decide on Board NIC.

8. Very Very important YOUR ESX host IP in KS script should be in the same IP range of appliance .

Some of the error message you will come across like

1. It will keep asking for keyboard type and various other question . This means you have not specify correct kick start device in your script. Please do specify correct KS device as I explained above.

My VMTM thread about it http://communities.vmware.com/message/1230094#1230094

2. PXE does not support 802.1q.

3. Also you can debug this appliance using ALT+F3 and ALT+F4. These you need to perform on ESX host which you are building. What I have learned that these are anaconda logs which help you to troubleshoot the appliance. You need to press ALT+F3 and ALT+F4 as soon as GUI method of install start.

4. Some of the error like

ALT+F3 :

Trying to mount device HDA

No network drivers for doing kickstart

Assert: No network device in chooseNetworkInterface

ALT+F4

cdrom:open failed

Such error message suggest that you have not selected correct KS device for your PXE.

5. If you are using EDA then make sure that you associates the ISO using CD ROM ISO. Which means you copy the ISO to some NFS or shares to connect to VM appliance. I tried connecting ISO using ISO connect ICON and it was keep failing. If you are using UDA then it is advisable to create separate disk of 20 to 40GB of drives and copy ISO over there.

6. If your ESX build stuck at the screen shown below and ALT+F3 shows “ *transferring http://10.0.0.104//esx/esx301//VMware/base/netstg2.img to a fd” .

“unable to retrive http://<ip of UDA>//esx/esx301/vmware/base/netstg2.img”

This means your UDA/EDA appliance are not in the same range of IP as of your ESX host getting build. You need to modify the KS script and ensure that your IP subnet is same as your ESX host getting build. That is the reason I suggested above to have IP reservation of your DHCP server

Some more ALT+F3 and ALT +F4 screen

For any further troubleshooting check the VMTM forum/Appliance website and their forum

For UDA http://communities.vmware.com/thread/43958?tstart=0 http://www.ultimatedeployment.org/forum http://www.ultimatedeployment.org/index.html

Person to contact _the_dude_ or Mike Laverick for any help on VMTM thread

For EDA http://communities.vmware.com/thread/139465?tstart=0 http://virtualappliances.eu/

Person to contact burgh on VMTM thread.

First Cut feature of ESX host 4.0

As I mention in my last note that I am installing vSphare ESX4.0 . Since I am not installing any vSphare CenterI decided to download the agent and use it for the time being to find out what all has been changed.

1, First change I noticed after installing vSphare agent that it allow you to login with current credential.

2. As soon as you login you get nice inventory panel to explore

3. Then I found very interesting thing. You can view logs (click on System Logs as shown above) under /var/log/vmware

4. You can do almost everything using this interface whereas earlier if you bypass virtual center server you have limited feature. One this I found that you have plugin available by Hyper9 and Andrew. Its same Andrew who has created storage vmotion plugin. Hyper9 give you nice little reflection and it is freely available on Hyper9 website. This plugin also work with VC2.5. One thing I also notice about spelling of Plugin, vmware spelled it as plug-in J

This is how your vm’s will look like if you have Hyper9 client installed. If it is windows host it will show MS symbol and if it is Linux it will show Linux symbol

5. They have improved the summary page

6. Console memory has increased from 272MB to 500MB as default.

7. Networking helps you to enable IPV6

8. Storage adapter also list your local ESX CD ROM . Not sure the reason but looks very interesting

9. New interesting feature called Advance setting. Not sure reason for this but looks very interesting

10. Licensing feature also looks very interesting

ESX4.0 install first CUT- Step by Step comparision with previous version

Publish at Scribd or explore others: