Octopus (Default) instance becomes unbound

I’m currently trailing and have successfully setup a CI + Octopus configuration. Earlier today, for some reason, the Octopus service shut down and now refused to start. It now tells me that is hasn’t been setup:

Instance OctopusServer of application OctopusServer has not been created. Check the instance name or run the setup wizard.
System.ArgumentException: Instance OctopusServer of application OctopusServer has not been created. Check the instance name or run the setup wizard.

So, I go to apparently re-setup the (Default) instance, but I can’t connect to the same database because it’s expecting me to provide it with a master key. Sadly, I can’t get to the master key from the UI because it thinks the (Default) instance isn’t installed, and if I run the following from PS Write-Host (.\Octopus.Server.exe show-master-key)[2] it fails because:

Error: The default instance of OctopusServer has not been created. Either pass --instance INSTANCENAME when invoking this command, or run the setup wizard.

There is nothing else in the event log to suggest why it has seemingly become unbound. The only thing which has occurred today was a disruption of our internet connection because our ISP was having some issues.

Any thoughts?

Hi Matthew,

Thanks for getting in touch and sorry for the issue you’re having.

This is an unusual issue - I haven’t seen it before! It looks like the Octopus configuration has either been corrupted or lost somehow.

Can you open regedit and navigate to HKEY_LOCAL_MACHINE\SOFTWARE\Octopus\OctopusServer and see what nodes are underneath that? There should hopefully be one there which has a string value pointing to a configuration file. Can you send me a screenshot of what you find there?

If there is a pointer to a config file, can you have a look inside to see if there’s anything that seems amiss? I’m not sure exactly what to look for, so if you find one, it might be easier to attach it so I can have a look.

If there’s no registry setting, have a look in the C:\Octopus folder (or whatever “home” folder was originally configured) to see if you can find a config file anyway.

Could anything have run on that machine that could have caused a registry cleanup?


Hi Damo,

I compared the registry hive on my evaluation box vs a new install on my current machine, and the OctopusServer subkey, i.e. HKLM/Software/Octopus/OctopusServer/OctopusServer was missing, so I’ve re-added it and added a string value ConfigurationFilePath pointing to my install location config (in this case C:\Octopus\OctopusServer.config) and this has corrected the issue. I’m not aware of anything that would remove registry keys, but our IT team are investigating this also.

So to summarise - fixed registry keys and service is running, but not idea why those keys were removed.

Hi Matthew,

Ok, that is very curious. I honestly have no idea how they would have been deleted either I’m sorry!

I’m glad you got it up and running again though. Let me know if it happens again.


Hi everybody,

this thread saved my life! :slight_smile:

I can give you a scenario how we face similary problem as Matthew. I was configuring the octopus server bindings (changing https certificates). When i clicked apply it gave me an exception: the disk is full. That was correct, the disk was full. The Octopus already deleted (or cleared) its config file (OctopusServer.config). When I started the Octopus server manager again it won’t start just show an exception that can’t parse az xml file. Then i deleted the empty configuration file and i think this was the time when the octopus server delete the registry keys.

First i copy and ol version of the config file back (fortunatelly we backup whole VMs weekly). But the Octopus server didn’t see any instances. After a heart attack i found this thread. I added the above mentioned registry values it started to work again.

In my opinion it is not good way to clear the whole config file (with the master key) before it is sure Octopus can replace that file.