Viewing the World in Binary

ESX 3.5 Host Update3 and more!

No more downloading packages, untaring and figuring out the sequence of install. We introduce Update Manager, another plugin from VMware to make our life easier. ESX host cluster update is a few clicks away. Lets start and document:

First step make sure the Virtual Center version and Update Manager are compatible versions:

comp-matrix

Begining with Virtual Center 2.5 and ESX Server Version 3.5 Update Manager can perform compliance scans and apply patches for guests, appliances and hosts.

As we will applying update 3 to ESX Server and Virtual Center, the following link will help us:

http://www.vmware.com/support/pubs/vi_pages/vi_pubs_35u2.html

First Step is to Upgrade Virtual Center and Update Manager to the latest release once that is done we Remediate ESX host.

Before Remediating the ESX host we need to attach baseline (and it is very much advisable not to upgrade the cluster at once). Select the host -> Update Manager tab -> “Attach Baseline”

attach-baseline

Check the Baseline you will like to Remediate the server with and click on “OK”.

The server did not come up after the Remediation process of Rebooting host and console showed it was stuck in Boot Selection mode. No problem! Selected exit using the always helpful IBM RSA and the server came back up. Looks like this delay for some reason lead to the VC declaring update was not successful specifically for ESX350-Update03.

um-fail2

Checked Error Log and saw the same message for ESX-Update03

failed-update35u32

I went ahead and checked the Baseline on the installed updates to make sure if 3.5Update 3 did actually fail. Baseline reported it as installed:

installed-update35u3

Paranoia gripped and I re-scanned the host and started another remediation process. It reported no non-compliance after the scan and remediation did not install any updates. I also logged in and checked using esxupdate -l command for the update, looks like alls well and its installed. Moving to another host on the cluster, the process went smoothly.

update-success

 Its clear that after the reboot ESX350-Update03 does complete its update process as we can see two “Install of update ESX350-Update03” messages in a appx. 10 minute interval.  Time to get done with this update and get to doing some Cisco stuff.

Advertisements

December 10, 2008 Posted by | ESX-VMware | | Leave a comment

ESX – Insuff. resources .. for HA

Got this error when trying to start a new virtual machine. Bad News!! Cluster is running is out of memory.

No Problem!! Check “Allow virtual machines to be powered on even if they violate availability constraints” and you are good to go. This setting is found under Cluser Settings -> ‘Vmware HA’. This is a quick dangerous way to start additional VM’s.

You really have to make sure that the ESX hosts are using less than 70% of memory (This is purely my guestimate!!), it can be much less % in certain scenarios where servers tend to take big chunks at intervals. Quick check on the host mem % in cluster can be made under cluster -> hosts.

How to estimate how many VM’s I can run on my clustered infrastructure?

Simple!

  • Take the VM-host which has been assigned the most memory e.g 1Gb. Figure out which ESX-host in the cluster has least amount of memory (if all ESX hosts in the cluser have same Gbs of memory, then you dont need to figure out any thing) e.g the number is 24Gb.  So 24/1 = 24, if you have three ESX-hosts in the cluser then multiply 24*3 = 72 VM-hosts. 

To power on another VM-host without a sweat and keep ‘violate availability constraint’ enabled,  add another ESX-host to the cluster.

October 24, 2008 Posted by | ESX-VMware | | Leave a comment

ESX – Orphaned VM

 

 

Login to the host where VM was last running.

  • vmkfstools -D /vmfs/volumes/datastore/<any dir to save>    // will dump the messages information in /var/log/vmkernel.
  • cat /var/log/vmkernel | grep owner

The above ‘cat’ command will throw something like ‘44444444-55555555-6666-00166666666‘, check the latest owner message.

The bold portion is the system uuid.

The following command will print out the system uuid when run on each of the ESX hosts:

  • esxcfg-info | grep -i ‘system uuid’ | awk -F ‘-‘ ‘{print $NF}’

Compare the ‘cat’ command output with ‘excfg’ command output, Identify host which is owner of the locked process. Run the following command on the identified host to list the process which is holding up the vm hostage:

  • ps -elf | grep <hostname i.e virtual machine name>

If you get back the process number which is holding up the file, kill the process and thats it.

This never worked for me :D, process state (ps) commad did not show any process with the ‘vmname’. I tried deleting the files from virtual center server console and it deleted everything except the vmdk file. A quick VMWARE Communities lookup indicated resoultion as rebooting the ESX host. Next day I tried deleting the directory and vmdk file, It Worked!! So there was some thing holding up the file and released it overnight.

October 22, 2008 Posted by | ESX-VMware | | Leave a comment