Cloudmin Pro license gets installed successfully, but no changes are seen.

Hi everyone.

We've recently replaced one of the worker nodes in our cluster, and after installing Cloudmin on the new server, we're seeing two issues.

The first issue is that when we use the "Upgrade to Cloudmin Pro" link and enter our serial number and license key, Cloudmin comes back with a message saying the upgrade was successful, even though there was an APT error during the upgrade. After refreshing the page, or logging out, or rebooting the server, or re-checking the Cloudmin license, I see no changes (I'm especially looking for the Virtualmin Replication feature), and the link to upgrade to Cloudmin Pro still exists. We have a 10 instance license, and I expect that would cover the two instances we have, even plus the defunct one on the old server.

Attached is the screenshot of he successful Cloudmin Pro upgrade. I wasn't able to reproduce the APT error though, so I'm not sure if that even matters.

The other issue we're having is related to virtualization. It's technically not even a feature we need yet, but it displays a prominent message on the System Information page. We get the error "This system cannot be a KVM host : Your CPU does not have the VMX or SVM extensions needed for KVM", in spite of the fact that all the virtualization options are turned on in the server's BIOS. We're using a Dell Poweredge 2950, and the configuration of this node is the same as the first one in our cluster, which coincidentally, has no trouble at all with either of these issues.

Status: 
Active

Comments

So did the Pro upgrade succeed the second time you tried it, or are you still stuck on the GPL version?

Regarding the VMX error message, can you post the /proc/cpuinfo file from your system? That is where Cloudmin looks for the vmx or svm flags.

No, it doesn't matter how many times I try to do the Pro upgrade, it never changes.

The flags from /proc/cpuinfo are exactly the same on worker1 and worker2. I don't see why one would have this warning about the KVM host, and the other wouldn't. I don't think I've turned that feature off, either.

Here's the Flags line out of /proc/cpuinfo:

flags : fpu de tsc msr pae mce cx8 apic sep mca cmov pat clflush acpi mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc rep_good nopl pni monitor est ssse3 cx16 sse4_1 xsave hypervisor lahf_lm dtherm

In that flags list, I don't see vmx or svm listed, which is what Cloudmin looks for to determine if KVM can be used. What kind of CPU is this?

It's a 2.83 GHz Intel Xeon E5440.

If we can't use KVM, how do we turn off the notification? I seem to recall that Xen would work fine anyway.

KVM isn't our main concern though, we care far more about making Cloudmin Pro work.

Yes, unfortunately, it looks like that particular CPU may not support KVM. Some have suggested that it may be possible to get this working by upgrading the BIOS, or enabling certain features in the BIOS, though that doesn't seem to have worked for everyone.

As far as upgrading Cloudmin goes -- can you attempt the upgrade again, and paste the output you receive during the upgrade process into the request here? Thanks!

Okay, here's an attachment with the complete output.

Ah, that output actually looks good -- are you saying it's still showing up as the GPL version though?

If so, what output does this command produce:

dpkg -l webmin-server-manager

root@worker2:/mailhome# dpkg -l webmin-server-manager

Desired=Unknown/Install/Remove/Purge/Hold

| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend

|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)

||/ Name Version Architecture Description

+++-===================-==============-==============-===========================================

ii webmin-server-manag 8.3 all Webmin module for 'Cloudmin'

That looks like you are running the Pro version - the GPL version has .gpl or .kvm in the version number.

If you're still seeing something that makes it look like you have the GPL version there, you could try restarting Webmin just to make sure something isn't cached.

I've rebooted the whole server several times trying to get this to work (please see my initial trouble report), but I suppose I could try that again.

Any other suggestions?

I've just rebooted the server again (it had a recent kernel upgrade anyway) without any changes.

But after I noticed that the new Virtualmin version in dpkg doesn't match the Virtualmin version in System Information, I figured I'd grep dpkg for that version:

root@worker2:/home/ernied# dpkg -l |grep 4.18.gpl ii webmin-virtual-server 4.18.gpl all Webmin module for 'Virtualmin Virtual Servers (GPL)'

I've removed then re-installed Webmin and all its dependencies, and that seems to have gotten it working.

At least, I think it has, because I've also been trying to get Postgresql working, and there were a few system changes there too, but more than likely they're completely unrelated.

I can get into the Cloudmin Replication settings in Cloudmin now, and that's what I'm the most concerned about.

Ok, if the replication settings are showing then you have definitely upgraded.

Ok, since this is a new server in the cluster, and I would expect that adding additional servers to the cluster would be the same process, I would totally expect that there should be an easy way to replicate Virtualmin's virtual servers to the new systems in the cluster.

However, simply checking off "All on source system" under "Virtualmin domains to replicate" doesn't do the job! If I select more than one domain to replicate, the process appears to complete properly without error, but that domain doesn't show up in Virtualmin on the new server. This only works if I select only one domain to replicate.

This is madness. I can't replicate hundreds or thousands of virtual servers one at a time. Please fix this feature ASAP and inform me of a new update!

What output do you get from the replication process if you select all or multiple domains?

Interesting addendum to that:

It appears to be a select few domains that are blocking this process, but they fail transparently. I've already found a couple that would say they're getting replicated, even one at a time, yet actually fail to show up on the second server. On the other hand, when I include domains that don't include those "broken" domains, I can replicate more than one at a time.

I'll do what I can to find out what's "broken" about the domains that won't copy, but at the same time, we still need them to replicate!

Can you backup these domains normally on the source system from within Virtualmin?

Do you mean back them up and restore them to the new secondary server?

That isn't going to work as well as replicating them, because of IP address conflicts. These IP addresses are already set up on worker1, we don't need them on worker2 as well.

One issue I've been able to isolate is that sub-servers won't replicate until their parents are completed. Again, the replication will simply fail silently without any errors.

No, I was just interested to know if backing up those domains works at all - if this fails, that would explain why the replication process isn't working, as it used backups and restores internally.