Nutanix Upgrade fails at 93%

Yesterday I did a plan upgrade of our 16-node (4 block) Nutanix 3050 cluster. The version which were running in last couple of months was 4.5.3.2 and the plan was to upgrade to 4.6.3. Using the GUI and the 1-click upgrade we took the following steps:

  1. I’ve Downloaded the upgrade binaries and metadata file from the Nutanix portal:1
  2. Went to the software upgrade section and uploaded the binaries there:243
  3. Wait for the upload to complete the I started the upgrade process:5 6 78
  4. The beginning was fine, but after few minutes I noticed that one of the CVM upgrade is failing at 93% and the process was restarting:10
  5. After this was repeated several times I realized that there is something wrong. Doing some troubleshooting I have realized that the SeviceVM ISO is not mounted to the VM’s CD-ROM. After checking with the Nutanix support, they confirmed that the ISO must be attached all the time:9
  6. It is very important to make sure that the operational procedures are updated and your scripts does not include any of the Nutanix Controller VMs and always follow the vendor recommendations.
The following two tabs change content below.

Nikolay Nikolov

VDI Engineer
Nikolay has 9 years work experience in IT and 5 of them in the Virtualization technologies mainly based on VMware products. Currently works as VDI Engineer at MSD IT Global Innovation Center and he is an ex-member of VMware CoE at IBM. He holds VCIX6-DCV, VCIX6-DTM and VCP on DCV, DTM, NV and Cloud, Nutanix NPP certificate and also Master Degree of Computer Systems and Networks. Honored with vExpert 2015/2016 by VMware and Nutanix Technology Champion 2016/2017.

Latest posts by Nikolay Nikolov (see all)

About Nikolay Nikolov

Nikolay has 9 years work experience in IT and 5 of them in the Virtualization technologies mainly based on VMware products. Currently works as VDI Engineer at MSD IT Global Innovation Center and he is an ex-member of VMware CoE at IBM. He holds VCIX6-DCV, VCIX6-DTM and VCP on DCV, DTM, NV and Cloud, Nutanix NPP certificate and also Master Degree of Computer Systems and Networks. Honored with vExpert 2015/2016 by VMware and Nutanix Technology Champion 2016/2017.

Bookmark the permalink.

3 Comments

  1. Good blog, two things to note:
    1). Never mess about with CVMs – Nutanix don’t even want you to snapshot them as it can cause IO issues.
    2). Notable that a cluster upgrade process could fail and no indication of any issues to the running systems.
    I’ve never had an upgrade fail, but it is comforting that even in failure Nutanix remains robust.

    • I completely agree with both of your points! The thing I wanted to describe is not that my upgrade failed, but that you must exclude all CVMs from any kind of scripts. The customers are running lots of scripts against vCenters and once a Nutanix cluster is added to those vCenters, the scripts must be changed, otherwise they will end up in my situation.

  2. Hi Nikolay,

    your blog is very good. Thank you!

    I have a little problem with a AppVolumes 2.11. Under Activities (Pending Actions, every day, I see 3/4 records like this: Audit VM state and ensure attachments are in sync – Failed 5 times”

    I think there’s a problem with a job named “Audit VMs”. But I don’t know where. can you help me?

    Thank you

    Francesco

Comments are closed