Jump to content
Welcome to our new Citrix community!
  • 0

Can't re-enable HA

Jon Berry


Have two servers in a pool, running XenServer 7.1.  We had a crash of the master, the mater failed over to the second server.  Took the old master out of HA and restored it to service in the pool.  Second server remains the master at this time.


In trying to re-enable HA, we are getting this error:


This operation cannot be performed because VDI OpaqueRef:b0fce4a4-9f68-2b74-3bce-6b2c05b2290d6 could not be properly attached to the VM.


Can anyone give me some guidance on where to start looking to resolve this?

Link to comment

Recommended Posts

Thanks for all your help.


So your thought is that this is a HOST UUID?  I do not believe we have removed any hosts since this was set up, only a VM.  There were three VMs in this pool, but now there are only two, but its the same two host nodes from the start,


The statefile.db editor does not want to work with 7.1 it seems, but I DI see the UUID in question in the statefile.db file, sadly alontg with a couple of other ones.


Editing this by hand gives me the heebie jeebies though.

Link to comment

Ejecting and removing the host will remove all of its references, so if its indeed host related in the xapi database

when you install it clean and re-add it the host uuid will be a new different on and won't give you those errors

anymore. I don't blame you for not wanting to edit that database directly, I have not had a need to do that 

either and would avoid it if at all possible. Just trying to come up with some ways to help resolve the issue.

Since you don't use loca SR's ejecting it from the pool, installing fresh, and re-adding it is the way to go.





Link to comment

The state file can be pushed using a "xe pool-sync-database" command which will force the master 's DB to propagate to other hosts.

I still agree that ejecting and re-adding the hosts should solve the bad state of whatever it happens to be in. Just be sure not to have any VMs on local storage or they will be destroyed in the process.


Your NTP times look fine. Again, with a two-host pool, I'd recommend looking into HA-Lizard if you want to run a working HA setup with just two hosts.



Link to comment

Sorry for the delay guys.  Been busy here.


Ok, so let me make sure I understand your suggestion.


Currently server A is Master and sever B is other node,


1.  Migrate VM on server B to server A.

2. Eject Server B from cluster

3. add Server B to cluser

4. Make server B the master

5. Migrate both VMs over to servver B

6. Eject server a from cluster

7. Add server a back to cluster

8. move one vm back to running on server a


Once complete, ideally, I should be able to re-enable HA?


Link to comment

If you eject a host from the pool and reinstall it and you use iSCSI be sure and make a not of the IQN so you can 

modify that after the rebuild unless you don't mind going into your SAN and updating the IQN.  I actually have

a list of customizations to keep me on track for server rebuilds.  But with just one host being a problem child 

kicking it out and re-adding it is a great idea.




Link to comment

Well, this turned into a whole thing, I ended up reinstalling Xenserver on each node in sequence because taking out out of the cluster broke the storage network and no amount of resetting it would get it back.  Very annoying.


Thanks for all the help guys!

Link to comment

Storage connectivity shouldn't have been totally destroyed as long as each node was being operated on separately, though in some cases, NFS and iSCSI storage networks on hosts do have to be re-created, depending on the circumstances. I usually take snapshots (with a camera) of the network configurations on each node before attempting such operations, just to have a good, current record.



Link to comment

The problem was with the private network itself, and even though UI reconfigured it the way it was, it would just not connect to the private network.  Instead of beating my head against  a wall with it, I just decided to reinstall Xenserver on each node after I took it out of the cluster.


Everything is 100% now again though.

Link to comment

I take it you tried an emergency host rest? If that doesn't work, I try a forced eject the host from the pool and then add it back in.  Try first

'xe host-forget host-uuid=HOST-UUID' and if not, 'xe pool-eject host-uuid=HOST-UUID' and maybe add a "--force" flag which can't really hurt.

Ejecting a host from a pool will reset the local SR, so save anything from it first elsewhere before you try that.



Link to comment

BTW I did restart toolstacks on both before posting to forum, that did nothing.  This post is the EXACT same problem but with no solution




This returned an empty result:


[root@ctx2 ~]# xe vdi-list uuid=b0fce4a4-9f68-2b74-3bce-6b2c05b2290d6 params=all
[root@ctx2 ~]#


Its like Xenserver thinks there is the VDI out there, but I cannot find anywhere where it is and have no idea how to remove it

Link to comment


This topic is now archived and is closed to further replies.

  • Create New...