Category: vSphere 7.0

VMware Tanzu HAProxy Troubleshooting

This blog post is dedicated to HAProxy Troubleshooting for vSphere with Tanzu or also known as TKGs. Based on your configuration and deployment and the various items you need to configure you can make mistakes or items are not correctly configured. In my case, there were multiple problems at different deployments with parameters and reachability related to the network. In the end, after all the hours of troubleshooting, I ended up with a list of commands that might help others out. So that is the topic of this blog post.

HAProxy Background

First an introduction about the product HAProxy. HAProxy is a load balancer that is used by vSphere with Tanzu. This is not mandatory but is a product to choose from. The main reason for HAProxy compared to the others is that it is completed free/open-source. The HAProxy OVA is packaged and delivered by VMware and can be found in the following repository. All commands below have been tested against the HAProxy v0.2.0 version (haproxy-v0.2.0.ova) that is at the moment of writing the most recent version available.

Appliance access (SSH)

After a successful deployment, you can access the HAProxy appliance with an SSH session. This session can be established with a tool like PuTTY. The user account that should be used in the root account.

Keep in mind: Do not change configuration unless you absolutely know what you are doing. Almost all the issues I ran into were related to entering incorrect information into the deployment wizard or firewall issues.

Troubleshooting Services

One of the first things to check at first is that all services are running on the HAProxy appliance. When services are not started this is mostly caused by an invalid/incomplete configuration that is filled by the deployment wizard of the OVA.

### Check failed services
systemctl list-units --state=failed

### Check primary services for HAProxy and Tanzu Integration
systemctl status anyip-routes.service
systemctl status haproxy.service

### Restart services
systemctl restart haproxy

Troubleshooting Configuration Files

There are multiple configuration files in use by HAProxy here are the most important ones. Also, keep in mind what I already said before… do not change anything unless…

### Anyip-routes configuration file
cat /etc/vmware/anyip-routes.cfg

### HAProxy configuration file
cat /etc/haproxy/haproxy.cfg

### HAProxy dataplane api configuration file
cat /etc/haproxy/dataplaneapi.cfg

### Validation of configuration file
haproxy -c -f /etc/haproxy/haproxy.cfg

Troubleshooting HAProxy process output

Sometimes it is good to check the latest messages generated by the HAProxy process. There will be information about the startup of the process and the pool members.

### Show logging
journalctl -u haproxy.service --since today --no-pager

Troubleshooting IP Settings

By entering wrong IP information in the deployment wizard the configuration files surrounding the IP address settings, gateway, etc can be configured incorrectly. What I noticed is there is not really a check inside the deployment that verifies if the address that is entered is valid in any sort of way.

### List IP Settings
ifconfig

### Config files (incase of three NIC configuration)
cat /etc/systemd/network/10-frontend.network
cat /etc/systemd/network/10-workload.network
cat /etc/systemd/network/10-management.network

### Routing check
route
ip route

Troubleshooting Certificates

Certificates files used by the HAProxy application are inside the HAProxy directory on the local system. The certificates are BASE-64 encoded!

### Certificate authority file:
cat /etc/haproxy/ca.crt

### Certificate server file:
cat /etc/haproxy/server.crt

### Certificate URL by default:
https://%HAProxy-Management-IP%:5556

Troubleshooting NTP

One of the all-time favorites that are notorious for disrupting IT systems is off course NTP. Here are some commands for troubleshooting on Photon OS.

### Check service status
systemctl status systemd-timesyncd

### Show NTP peers
ntpq -p

### Restart service
systemctl restart systemd-timesyncd

### Configuration file
cat /etc/systemd/timesyncd.conf

Troubleshooting the HAProxy API

The HAProxy API is used by Tanzu to configure HAProxy for the management and workload components. Authentication is set up when deploying the OVA and the credentials are entered in the wizard. With the second URL you can verify those credentials:

### Info page
https://%IP-address%:5556/v2/info

### Authentication should work with the HAProxy user account (specified in the deployment wizard)
https://%IP-address%:5556/v2/cluster

Wrapup

Thank you for reading this blog post about HAProxy troubleshooting for vSphere with Tanzu or in short TKGs. I hope it was useful to you! If you got something to add? Have additional tips or remarks please respond in the comment section below.

Have a nice day and see you next time.

Source

HPE ProLiant DL20 Gen9 SATADOM Installation

Today we are going to work on an HPE ProLiant DL20 Gen9 server. After the initial installation, I was using an SD card as boot media but I still had some Delock SATADOMs laying around from my older lab servers that were replaced. So it was time to improve the performance of the boot media in the servers. In this blog post, I am explaining in detail the SATADOM installation in an HPE ProLiant DL20 Gen9.

So what are the advantages compared to an SD card:

  • VMware ESXi boot time about 50% faster
  • VMware ESXi upgrade time about 70% faster
  • Inventory performance (very noticeable when clicking through the VMware vCenter or VMware ESXi web GUI)
  • The overall stability of the host, this because of the “high” failure rate of the SD card.

The summary of advantages is based on my own comparison between SD cards en SATADOMs in my ESXi Hosts in my Home Lab.

Delock SATADOM Specifications

Here are the specifications of the Delock Satadom devices I am using for both HPE ProLiant DL20 Gen9 servers. Here are some tips about what I have learned so far… I bought them in 2018 so they are not brand new anymore:

  • Buy them a little bit bigger because of the future proof > minimal 32GB I would suggest.
  • Verify before buying if you need the vertical or horizontal model (rack model server go for horizontal / tower model server no really important).

So here are the specifications from the Delock website:

ItemValue
VendorDelock
TypeSATA 6 Gb/s Flash Module 16 GB vertical
Part nr54655
Capacity16 GB
InterfaceSATA 6 Gb/s, SATA 3 Gb/s, SATA 1.5 Gb/s
Performance460 MB/s read – 160 MB/s write
Power usage1.0 W max. (5V x 200mA)

SATADOM Installation

So now it is time to install the device on the server. Of course, it is a little more complicated in a small half-size rack server. For example, there are no Molex power connections available by default. So in the end the cable kit is almost more expensive than the device itself. The preferred option should be to find an HPE cable kit, not sure which one you will need. So after some thinking and looking into the server I came up with the following solution to just plugin the SATADOM.

At first, I needed to find a SATA port on the motherboard. Both ports are available in my case but I used the one that is normally used for the DVD ROM drive number 14 (see the image from the HPE manual).

The storage device itself can be placed in the space of the storage controller battery pack. Both of my machines do not have the expensive storage controller option. Only the onboard default controller. So the space is completely empty and an easily accessible location for the SATADOM.

The power is the most difficult one. I ended up with converters to into the power connection from the storage backplane (keep in mind my server has no internal storage except the boot device (the SATADOM in this post…) If you have your storage filled with SSDs or HDDs you need to figure out a new solution where to get the power from. I have read something about a power kit for the DVD ROM for example. I have never seen it on a picture or in a server so I do not know which connectors are in that cable kit but it might be an option.

To make some more sense and pictures explain more than words… Here is a gallery with some pictures of the SATADOM installation:

DL20 Gen 9 BIOS Settings

After the physical installation, it was time to set up the BIOS. To be honest it was quite easy compared to the HPE Gen8 where I had a lot of problems because of the ports and bios settings.

Here are two screenshots. The first one is the activation of the internal storage controller. Note: make sure you power cycle the machine before the SATADOM is detected. After the power cycle, the VMware ESXi installer should detect the SATADOM when trying to install VMware ESXi.

After this point, the SATADOM installation is completed. Just continue your normal procedures and put your host into production when you are done.

Wrap-up

So that is it for today…! I hope it was useful for other people and interesting to read. Keep in mind this blog post was focused on the HPE ProLiant DL20 Gen9 but I think the procedure will be quite identical to other HPE Gen9 servers. The most difficult part will always be the cabling and after that, the BIOS settings to get the device detected correctly.

So far my hosts have been running for about 40+ days without any issues and are working perfectly fine. If you got additional questions or remarks please respond in the comment section below. Thanks for reading my blog post and see you next time.

HPE ProLiant DL20 Gen9 Home Lab

This blog post is about replacing my current 24×7 Lab with a new set of two HPE ProLiant DL20 Gen9 servers. In this blog post, I am going to tell you about the configuration of the machines and how they are running on VMware ESXi. Also, I am going to compare them to my other lab hardware and my past home lab equipment.

Hardware

So let’s kick off with the hardware! The HPE DL20 Gen 9 servers I bought were both new in the box from eBay and I changed the hardware components to my own liking.

A couple of interesting points I learned so far nearly all servers that you will find for sale are provided with an Intel Xeon E3-12XX v5 processor. One item you need to take into account: yes you can swap the CPU from a v5 to a v6 like I did but you need to replace the memory modules also! The memory modules are compatible with a v5 or v6 processor but not both ways. The Intel Xeon E3-12XX v5 CPUs are using 2133 MHz memory and the Intel Xeon E3-12XX v6 CPUs are using 2400 MHz memory. So keep that in mind when swapping the processor and/or buying memory.

In the end, after some swapping of components, I ended up with the following configuration. Both ProLiant servers have an equal configuration (like it should be in a vSphere cluster):

ComponentItem
Vendor:HPE
Model:DL20 Gen9
CPU:Intel® Xeon® Processor E3-1230 v6
Memory:64GB DDR4 ECC (4 x 16GB UDIMM @2400MHz)
Storage:32GB SD card on the motherboard
Storage controller:All disabled
Network card(s):HPE Ethernet 1Gb 2-port 332i Network Adapter
Expansion card(s):HPE 361T Dual-Port 2x Gigabit-LAN PCIe x4
Rackmount kit:HPE 1U Short Friction Rail Kit

Power usage

So far I have measured the power usage of the machines individually with the listed configuration in the hardware section. When measuring the power usage the machine was running VMware ESXi and on top of about seven virtual machines that were using about 30% of the total compacity. I was quite amazed by the low power consumption of 31.7 watts per host but I have to take into account that this is only the compute part! The hosts are not responsible for storage. Here is a photo of my power meter when performing the test:

Screenshot(s)

Here are some screenshot(s) of the servers running in my home lab environment and running some virtual machine workload:

  • Screenshot 01: Is displaying one of the hosts running VMware ESXi 6.7 (screenshot from HPE iLO).
  • Screenshot 02: Is displaying one of the hosts connected to VMware vCenter and running virtual machines.
  • Screenshot 03: Is displaying one of the hosts HPE iLO web page.

Positives & Negatives

To sum up, my experience I have created a list of positives and negatives to give you some insight into the HPE ProLiant DL20 Gen9 as a home lab server.

Positives:

  • A lot of CPU power compared to my previous ESXi hosts, link to the previous setup.
  • Rack-mounted servers (half-size deep with sliding rails).
  • Out of band management by default (HPE iLO).
  • Power usage is good for the amount of compute power delivered.
  • No additional drivers are required for VMware ESXi to run.
  • The HPE DL20 Gen9 has been on the VMware HCL, link.

Negatives:

  • Noisy compared to my previous setup (HPE ProLiant ML10 Gen8). For comparison, the HPE ProLiant DL360 Gen8 is in most cases “quiet” compared to the HPE ProLiant DL20 Gen9.
  • Would be nice if there was support for more memory because you can never have enough of that in a virtualization environment ;).

Photos

Here are some photos of the physical hardware and the internals, I did not take any pictures of the hardware when the components were all installed. I am sorry :(.

  • Screenshot 01 – Is displaying both machines running and installed in the 19″ server rack.
  • Screenshot 02 – Is displaying the internals of the DL20 Gen9. Keep in mind this one is empty. As you can see in that picture the chassis is just half-size!

Wrap-up

So that concludes my blog post. If you got additional questions or remarks please respond in the comment section below. Thanks for reading my blog post and see you next time.