Yellow Bricks

by Duncan Epping

vcsa

Command-line upgrade of vCenter fails with Test transaction failed to update packages error

· ·

I have various test environments, and one of my environments I was testing the command-line upgrade of vCenter Server. Now, most of my environments we tend to use for destructive testing and strange update/upgrade scenarios, so we hit some strange issues every now and then. While I was doing the command-line based upgrade of vCenter Server (see this post for how to do that), I hit an error. The error was the following:

Test transaction failed to update packages

I looked at the log called “/var/log/vmware/applmgmt/software-packages.log”, and I noticed the following entry:

eventlog is obsoleted by (installed) syslog-ng-3.17.2-1.ph3.x86_64

I removed the package manually as follows:

rpm -e syslog-ng-3.17.2-1.ph3.x86_64

I then retried the update and it worked, as shown in the screenshot below!

Updating VCSA and hitting the error Test RPM transaction failed

· ·

I was updating my environment to vCenter Server 7.0 U2c, while going through the process I got this error that says “Test RPM transaction failed”. Below is the screenshot of the error. if you click “resume” you then, unfortunately, get stuck in an infinite loop. The only way to get out of the loop is by removing a file via SSH on vCSA called “/etc/applmgmt/appliance/software_update_state.conf”.

Updating VCSA and hitting the error Test RPM transaction failed

So now what if you want to update? We resolved it as follows, and let me include the deletion of the state file as well:

rm /etc/applmgmt/appliance/software_update_state.conf

Then we rebooted the VCSA:

reboot

Then we went into the appliance shell via SSH and ran the installer from the appliance:

appliancesh
software-packages install --url --acceptEulas

After which the installation was completed correctly.

Very large sps-runtime.log.stderr log file

· ·

On VMTN someone hit a situation where the sps-runtime.log.stderr log file grew extremely large on their vCenter Server Appliance (7.0 in this case). I have seen this before, and sizes over 10GB are not uncommon. The sps-runtime.log.stderr file belongs to the service that provides the Policy-Based Storage capabilities. You can of course stop the service and then delete the files, and restart the service again. However, you could also empty the file by simply doing the following on the command-line:

cat /dev/null > /storage/log/vmware/vmware-sps/sps-runtime.log.stderr

 

This results in a 0kb file immediately.

Unable to query vSphere health information. Check vSphere Client logs for details.

· ·

After an upgrade from 6.5 U1 to 6.7 U1 a customer received the following error in vCenter: Unable to query vSphere health information. Check vSphere Client logs for details. They looked at the log files but couldn’t get an indication of what was wrong. In this case, it was pretty simple, one of the required services wasn’t started for whatever reason. You can verify this in the vCenter Appliance VAMI (management interface for the appliance), which can be accessed by going to “http://ip-of-vcenter:5480”. When logged in you have to check the Services section, and make sure the VMware Analytics Services is running, as shown in the screenshot below.

Deploy VCSA 6.0 firstboot error

· ·

I was doing some tests in my lab and while deploying a new VCSA 6.0 I received an error that firstboot was unsuccessful. Not really a great error message if you ask me but okay. I had already validated DNS twice before I got started, but I checked it again just in case… DNS was all good, what else could it be? Figured NTP could be another problem and my friend William Lam confirmed that. I checked the host if NTP was configured and it was not for some reason. So I configured NTP on my ESXi hosts which was straight forward, but what about the VCSA I had deployed? Also not too complicated, I logged in via SSH and did the following:

  • ntp.get
    Will show “Status: Down”
  • ntp.server.add –servers 10.17.0.1
    This configures VCSA to fetch the time from ntp server to 10.17.0.1
  • timesync.set –mode NTP
    Make sure that the time sync is set to ntp
  • ntp.get
    Should show “Status: Up”

That should do it… By the way, you can simply check “resolv.conf” for DNS to see how it is configured today, also look at “hosts” for the host name etc.