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

Migrate using second host?


Jonis Maurin Ceará

Question

Hi.

I'm having a big problem here. I have XenServer 7.5 running with 3 hosts and an FB storage attached to this pool. I've created a new pool with 5 servers and another storage using iSCSI.

The problem is that in the first pool (P1) this storage is showing "7.5 Tb used of 8.2Tb total (14.1 Tb allocated)" and because of this (I think), pool master can't do anything in this storage.

Member 2 and member 3 connects and start any VM fine (VM's storage in this same storage), so it's working.... even master member of P1 can connect to storage (path), but can't do any commando or operation, even snapshot or export.....always got errors related to python in SMLog.....and something related to 'can't get physical size' in SMLog. But the thing is that this error just happen on master of pool.....all other members are connected fine to the same SR. And even if I change the pool master, the new master show the same error as old master.....and the old master working as second/third members now works fine starting VM's, so it's not related to host itself, but master of pool.

 

I'm using the latest patchs (updates) and still same problem. I tried to 'delete' some ghost disks/snapshots from this storage using 'xe vdi-destroy', but didn't work, got same error os SR:

Quote

 

[root@Kawasaki ~]# xe vdi-destroy uuid=b4f3ba90-967c-422d-ac62-f8ef9c9cf6fd
The SR has no attached PBDs
sr: d953467e-fb15-cc49-4eb5-3288d3ed2c82 (Magali)

 

 

 

 

So my question is: is there any way to export/migrate/do-anything using a secong pool member instead of pool master?  What else can I do to test/solve this? any ideas?

 

Tks!

 

Link to comment

5 answers to this question

Recommended Posts

  • 0

All such functions get pushed t the pool master, so you need to find out why this is happening and fix it.

 

I assume you've done the usual toolstack restart, sync the pool database, reboot, etc., right?  If you do an "xe sr-list params=all
 do you see a PBD associated with that SR? Looks like it may have lost its connection.  This is the fiber channel SR on which you see the issue, correct?

If you have any VMs on any other storage (such as local) do such operations work?

 

-=Tobias

Link to comment
  • 0

I typically run less VM's on my pool master just for this reason since it is in charge of pool wide operations.

What you can do is designate one of the other servers to be your pool master, move your VM's and reinstall

that host if need be. Your storage needs addressed. You can probably only move small VM's since most of

your storage is used.

 

--Alan--

 

 

Link to comment
  • 0

Hi Tobias!

Yes, done all possible tests....including shutting down everything and starting again (remove power plug, storage inclusive).

Yes, I have other storages (iSCSI in this case) and everything works fine with them. The thing that is making me crazy os that only master has errors, despiste which host is master. If I start any VM that is in this problematic storage, but start in other member that is not master, VM starts perfectly! Just don't start in pool master. Same for snapshoot, export, etc....

 

Looking at SMLogs I found some erros, specially this one:

failed to get physical size: -22

 

I tried to run command by hand and got same error.....so my guess is that vhd-util is not able to find storage size.....maybe because of that '14.1 Tb allocated'? I don't know.

[root@Ducati tmp]# vhd-util query -s -n /dev/VG_XenStorage-d953467e-fb15-cc49-4eb5-3288d3ed2c82/VHD-fea92b1a-a203-4895-ad2a-e5898d20c071
failed to get physical size: -22

Using XenOrchestra I could see a lot of orphaned disks, without VM.....and some orphaned snapshoots....I tried to delete using vdi-destroy, but no success....also, I tried to find physical files, but I could not find any mount point for my FC storage, nothing related...my idea was to delete those files manually (rm) using their UUID's as identification, and not 'vdi-destroy'.

But no success...

 

Link to comment
  • 0

Looks like you might have some major storage corruption or if nothing else, the SR is disconnected somehow. Does it show up as a red icon on XenCenter?

Have you tried re-plugging that SR? If you do that list command above, do you see an attached PBD? If the remote storage isn't working right, it won't help designating a new pool master.

 

-=Tobisa

Link to comment
  • 0
23 minutes ago, Tobias Kreidl said:

Looks like you might have some major storage corruption or if nothing else, the SR is disconnected somehow. Does it show up as a red icon on XenCenter?

Have you tried re-plugging that SR? If you do that list command above, do you see an attached PBD? If the remote storage isn't working right, it won't help designating a new pool master.

 

-=Tobisa

Tobias, 

Thats listed in XenCenter with a red X meaning error......but 'path' are all active, for all servers. I'm not sure if is a storage error, since all VM's starts fine in all hosts.

 

Let me try to explain again in a easy way:

 

Host 1: Ducati (master)

Host 2: Kawasaki

Host 3: Triumph

Storage: Magali

 

In this configuration, Magali is with "X" in Xencenter, multipathing is enabled and ALL servers has 2 of 2 paths active.

Storage "State" is "broken", with Kawasaki and Triumph "Connected", but "Ducati" as "Unplugged". Trying to do a 'repair' results in errors (already reported here) in SMLog. All VM's start and work fine in Kawasaki and Triumph host, no exception, everything ok (but I still can't do snapshot/export).

 

Now, make Kawasaki as pool master:

Host 2: Kawasaki (master)

Host 1: Ducati 

Host 3: Triumph

 

Now, storage "Magali" get connected status in both Ducati and Triumph, but 'Unplugged' in "Kawasaki"......and again, I can start/stop any VM in Ducati and Triumph, but not in Kawasaki.

 

This happens in any combination of servers.....master can't do any operation in storage, not even start a VM......but members can (start VM's, at least).

 

If this was a storage problem, I wouldn't be able to use or start any VM that are there :/

 

What am I trying to do is move all these VM's to a new storage, with more space.....and that's why I'm failing, since only master can do this such thing :(

 

 

 

 

 

 

 

 

 

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...