Nutanix AHV

Troubleshooting Common Nutanix CE Issues

TROUBLESHOOTING COMMON NUTANIX CE ISSUES

Disk detection issues and workarounds on Bare Metal

One of the more common issues during installation are disk detection issues. An error message
similar to the following is displayed:

The list of disks is either incomplete or empty or some entries are incorrect.
The most common causes are:

● SSDs are not recognized as such and are treated as disks. As a result, the installation
Fails because it requires at least one SSD.
● The RAID controllers are not set up properly (for example in a RAID 1 configuration
Instead of JBOD) or are not supported by AOS. As a result, the disks are not
Recognized. Vdisk or raid should be properly configured in BIOS
● Bad partitioning, with too small partitions, or left bits from other installations.
● The disks or SSDs actually do not meet the requirement of minimum criteria

To debug the issue, check what the installation environment actually sees, by first running the
following Linux system commands:


lsblk
lscsi
fdisk -l

These commands should give you some overview on how your disks and SSDs are seen by the
system. If any devices are missing, the most likely culprits are RAID settings or missing drivers
for your hardware. If the devices are listed properly, you need to check what the Nutanix
installation sees. Login as the root user with the default password and run the following
command:


/home/install/phx_iso/phoenix/collect_disk_info.py | json_reformat


The following example shows some basic data we can discern from the output:

In the above example, two block devices are found – one SSD as seen by rotational set to
false and isSSD to true, under /dev/sda; you can also recognize the model and the size.
Similarly, the second disk under /dev/sdb is a hard drive and the flags are correct. Both drives
also meet the size requirements.


When checking the output of the command, check if all devices are listed and their entries are
correct, for example the rotational and isSSD parameters, or if there are any duplicated
WWNs.

Disk type is not detected correctly One of the possible issues is that the SSD is mistaken for a HDD. To verify this, run: cat /sys/block//queue/rotational

Rotational should be 0 for the SSD and 1 for HDD. If all drives return 1, the system believes
there is no SSD and the requirements are not met. To fix this, set rotational to:

echo 0 > /sys/block//queue/rotational

Performance issues


Depending on some RAID controller settings, the IOPS results may be too low to meet the
default IOPS thresholds. If your setup will not be very demanding, you can lower them, for
details see the virtualization section Changing IOPS thresholds.

Duplicated WWNs
RAID controllers that reuse the numbers usually cause the issue. Check and fix the controller
settings

Missing firmware


If a particular firmware is missing from the installation image, but is available for download, you
may be able to add it to the image. The following example shows how the issue is detected and
resolved for the bnx2-mips driver. You may need to perform similar steps for other firmware.

Example

During installation, when starting the CVM, the following error indicates that the bnx2 firmware is
missing:

Can’t load firmware file “bnx2/bnx2-mips-09-6.2.1b.fw”

Sometimes, the network is not working properly and the same error message displays when you
run:

ifconfig eno1 up

The solution is to load the driver manually, from the command line. Search for and then
download the required bnx2-mips firmware and copy it to a USB key.

Login as root, mount the key, copy the firmware to the firmware folder, and restart the network:

mkdir -p /media/USB
mount -t vfat /dev/ /media/USB
cd /media/USB/
cp bnx2-mips-09-6.2.1b.fw /lib/firmware/bnx2/bnx2-mips-09-6.2.1b.fw
systemctl restart network

Also, intel virtualization technology must be enabled in bios for installation of Nutanix community edition.

Leave a Reply