After upgrading to VC 2.0.1 one of my ESX servers disconnected and after trying to reconnect I get the following errors:
Failed to install the VirtualCenter Agent service.
Login failed due to a bad username or password.
Also, I logged in to the ESX server directly and I see the following errors:
A general system error occured: Create upgrade tmp directory failed.
The specified key, name, or identifier already exists.
You could manually try to install the agent as follows:
- Copy the vpx-upgrade-esx-6-linux-32042 from the upgrade folder in the VC folder to /tmp on the ESX server
- run the command sh vpx-upgrade-esx-6-linux-32042
(By default the script is located in C:\program files\vmware\vmware virtualcenter\upgrade )
Next restart management services:
service mgmt-vmware restart
After completing the above steps, try to reconnect to the esx
I experienced the exact same issue - 4 ESX servers in one cluster, upgraded to 3.0.1 and everything was OK until I upgraded VC to 2.0.1. At this point, all of the servers were disconnected and wouldn't reconnect. The process outlined by trfa above fixed all 4. Thank you so much!
That worked. I was successfully able to reconnect to the VC. Thank you for your help
Thanks tfra.
I had the same situation and your solution resolved it for me as well.
Just an update for folks not using ESX 3 yet.
Look in C:\Program Files\VMware\VMware VirtualCenter 2.0\upgrade\bundleversion.xml to determine which vpx-upgrade-esx script you need. I had this problem and needed #1 for an old 2.1.2 system.
And the path to the script on my VC 2.0.1 load is C:\Program Files\VMware\VMware VirtualCenter 2.0\upgrade\
Message was edited by:
Jae Ellers
going from esx3.0 to 3.01 worked for me with upgrade 6.
I am having this problem with 1 out of our 38 ESX 2.5.3 hosts. We upgraded to VC2.0.1 from 1.3.1 last week and this 1 host won't connect with this error. I've tried a manual install using
sh /tmp/vpx-upgrade-esx-3-linux-32042
But to no avail. Don't really want to reboot the host as it's running VMs. Am going to log support issue and see what they say.
kim
Did you fix this already? I'm having same problem now .. after upgrading VC to 2.1 and ESX to 3.01
I ended up rebooting the host
Did you try restarting serverd?
DB
In my case I had filled up the / disk volume with ISO's !
I had to update the vmimages soft link to point to VMFS and move all the ISO files over to vmfs.
J
We just applied Patch 1 to VCenter 2.0.1 and experienced a similar issue in that our ESX 3.0.1 hosts configured for HA/DRS wouldn't reconnect to VCenter. Got the same "Failed to install the VirtualCenter Agent service." error as the first post. A standalone 3.0.1 host reconnected just fine as well as all of our 2.5.3 hosts. Fixed it by running
service mgmt-vmware restart
on each host that was having problems and then manually reconnecting via VCenter.
Message was edited by:
pjgould
I ran 'service mgmt-vmware restart' and it fixed it for me. Thanks!
Encountered this same issue when I installed VC 2.0.1 Patch 1 today. The service mgmt-vmware restart resolved it for some of my hosts, others who would connect but not register with HA required me to destroy the cluster and re-add the hosts.
Hi!
I had the same issue today.
Destroying the cluster and reastablishing a new one solved the problem.
This works, but isn't a real solution. What if there are realy sensitive VMs running on the ESX-Hosts? What if they are "mission critical"?
Disconnected VMs or even ESX-Hosts are real pain in the ass, at least for me
Regards,
Philipp
It works , thanks for your help Mr. tfra......
I had a similar problem when attempting to apply patch 2 to VirtualCenter. All but 1 of the ESX servers updated successfully. I was naive and did not realize that the VirtualCenter patch would also update my ESX servers. I determined that I was out of space on the /opt filesystem. Clearing space and restarting mgmt-vmware and vmware-vpxa did not seem to help. I was able to install vpx-upgrade-esx-6-linux-40644 manually, but was still unable to connect the ESX server in VC. What finally worked for me was to resynchronize the time with ntp, uninstall vpx manually, remove the vpxuser userid, and reattempt to reconnect in VC. BINGO!
trfa, your solution work for me after upgrading to VC 2.01. Thanks for the post.
Hello all,
My problem seems to reach further. I did all of the above. But the installation of "sh ./vpx-upgrade-esx-6-linux-40664" hangs (See log below). There are six VM's running on this ESX 3.0.1 server.
The VIC's error is Failed to install the VirtualCenter Agent service. Some help would be appreciated.
Total of 8 DL380 G5's with ESX 3.0.1 and only (this) one has this problem.
Now there is no vpxa installed anymore vpxa -v: -bash: vpxa: command not found. Uh, yikes!
NOTE: The root password is correct but VIC reports: Login failed due to a bad username or password. And then it starts to install the update and hangs for about two! hours on 19%, Reconnect host.
Log:
27747 2007-03-13 11:20:03 AM: Logging to /var/log/vmware/vpx-ivpx-upgrade-esx-6-linux-40644.log
27747 2007-03-13 11:20:03 AM: exec mkdir -p /tmp/VMware-vpx-esx-6-linux-40644
27747 2007-03-13 11:20:03 AM: status = 0
27747 2007-03-13 11:20:03 AM: exec tar -xvf /tmp/VMware-vpx-esx-6-linux-40644.tar -C /tmp/VMware-vpx-esx-6-linux-40644
27747 2007-03-13 11:20:03 AM: status = 0
27747 2007-03-13 11:20:03 AM: exec cd /tmp/VMware-vpx-esx-6-linux-40644/
27747 2007-03-13 11:20:03 AM: status = 0
27747 2007-03-13 11:20:03 AM: Build: 39823 List: LGTOaama-5.1.2-1.i386.rpm LGTOaamvm-5.1.2-1.i386.rpm VMware-vpxa-2.0.1-40644.i386.rpm
27747 2007-03-13 11:20:03 AM: Rpmlist: VMware-vpxa-2.0.1-40644.i386.rpm LGTOaamvm-5.1.2-1.i386.rpm LGTOaama-5.1.2-1.i386.rpm
27747 2007-03-13 11:20:03 AM: exec rpm -ev --allmatches VMware-vpxa
27747 2007-03-13 11:20:03 AM: status = 0
27747 2007-03-13 11:20:03 AM: exec rpm -ev --allmatches LGTOaamvm
TRFA's idea worked for us too.
ie : service mgmt-vmware restart
Many Thanks 
This thread is a gem
Two thumbs up from a poor sucker upgrading legacy ESX servers after midnight far away from home.
I'm having this issue with one ESX 2.5.3 host after applying the VC 2.0.1 patch 2 upgrade.
When I try and install the client upgrade, manually or from VC, I get this same error in the log:
20166 2007-04-11 11:20:58: Error installing VMware-vpxa-2.0.1-40644.i386.rpm: 1
20166 2007-04-11 11:20:58: status: 1. Cleaning up...
20166 2007-04-11 11:20:58: exec rpm -ev --allmatches VMware-vpxa
error: cannot get exclusive lock on /var/lib/rpm/Packages
error: cannot open Packages index using db3 - Operation not permitted (1)
error: cannot open /var/lib/rpm/packages.rpm
20166 2007-04-11 11:20:58: status = 255
20166 2007-04-11 11:20:58: All done! Status: 1
Any ideas?
Thanks!
Can you reboot your host?
Not easily. I was hoping to not have to.
Understood.
I don't know what the issue but I thought perhaps if you had VMotion and an additional host, you could bounce the box and try again.
If the host has become unstable, an outage may be in order anyway.
Easier workaround than installing the agent manually is to :
'mkdir /tmp/vmware-root' on your host and then try and add it to VC.
I had this error and discovered the rpm database on the host was inconsistent
Try doing an "rpm -qa" from the console and ensure no errors are displayed.
use "rpm --rebuilddb" to correct any errors
Then try reconnecting the host from the VI client
It worked for me
Thanks for the idea. I'd tried that with no luck. A reboot finally cleared up this issue.
Randy
After much research on the VMTN I just completed correcting this problem.
VC 2.0.1 patch 2, ESX 3.0.1 host would not connect.
Error message: "Failed to install the VirtualCenter agent service"
Restarting the agents did not solve my problem:
service mgmt-vmware restart
service vmware-vmkauthd restart
My process to manually install the agent (documented here and elsewhere) on VMTN was:
1) copy the file vpx-upgrade-esx-6-linux-40644 into /tmp on the host console. This file is stored in the VC directory "C:\Program Files\VMware\VMware VirtualCenter 2.0\upgrade"
2) execute the file with the command:
sh vpx-upgrade-esx-6-linux-40644
3) restart the management services:
service mgmt-vmware restart
4) delete any failed host connection entries out of the VirtualCenter
5) connect to the host using the VirtualCenter
Gene
After upgrading to VC 2.0.1 Patch2 two of our three ESX 3.0.1 build 39823 were disconnected in VirtualCenter.
Restart the agent
service mgmt-vmware restart
works fine!
At one of the ESX I have also to restart
service vmware-vmkauthd restart
Many thanks.
HB
After upgrading to VC 2.0.1-40644 the ESX servers disconnected and after trying to reconnect I get the following errors:
Failed to install the VirtualCenter Agent service.
Login failed due to a bad username or password.
Also, I logged in to the ESX server directly and I see the following errors:
A general system error occured: Create upgrade tmp directory failed.
The specified key, name, or identifier already exists.
my problem was solved only with
in etc/init.d/
service mgmt-vmware restart
and then reconnect the esx server in Virtual Center and everything works fine.
Just a restart to:
service mgmt-vmware restart
was needed here..
I disabled HA on the cluster connected the hosts back and enabled HA and that worked for me with the install of the script and all.
I have an ESX 3.0.1 install from months ago that is fully patched up till now and I have just done a fresh install of VC 2.0.1 with vpxd.exe version of v4.0.2.40644
The error is still occurs 
I used the:
service mgmt-vmware restart
after resorting to the forum for advice.
Thanks guys for your advice.
Running 'service mgmt-vmware restart 'resolved the problem but it restarted all my VMs running on the host.
Is there a way to prevent this? (This is a stand alone box, so I can't use vmotion)
Thanks
Carl
Message was edited by:
cbuchmann
I also have same issue, but I didn't upgraed the VC, but removed the old one 2.0.1 and install the new one 2.0.2. After that I could able to add 2 ESX host to the VC but when I tried to add the third one I got this message.
I was trying to find the file vpx-upgrade-esx-6-linux-32042 but I couldn't. I found vpx-upgrade-esx-6-linux-50618 instead.. Will this file work for me?
Help is really appreciated.
Thanks,
Ketal
ketalparikh,
I just went through this using the vpx-upgrade-esx-6-linux-50618 file that you mentioned, and it works.
I'll second that.
A host in Munich didn't like the upgrade.
The service mgmt-vmware restart worked just fine.
Thank you for your help!
I had 18 hosts saying "Failed to install the VirtualCenter Agent service."
service mgmt-vmware restart worked like a charm. Thank you so much.
I did the upgrade from VC 2.0.1 to VC 2.0.2 and had 2 of my eight Host servers not reconnect properly and give this error. I did the service mgmt-vmware restart and it resolved the issues. I also had to reconfigure HA on all of the servers before they would clear up completely. Thanks for the thread
One of my four servers did not reconnect after the 2.0.2 Update 1 for VC.
I issued 'service mgmt-vmware restart' on the host and could then reconnect to it.
restarting the service did the trick.
thanks
Yeah, the 'service mgmt-vmware restart' worked for me too. Thanks. When it doubt, reboot, I guess.
This document was generated from the following thread: Failed to install the virtualcenter Agent service after upgrading toVC2.0.1
