Reclaim free space xenserver 7.6

[scott budworth]

Hi, Guys!. 

 

i know this has been asked over and over but i cant find any information relevnt to xenserver 7.6

 

i have deleted old snapshots but the space has not been reclaimed by the system, should be aprox 2TB free'd up from snapshots. 

 

reclaimed space doesnt work on this version and i checked i have a LVM filesystem so colance should be running automaticly. 

 

4060.3 GB used of 5538.5 GB total (19920 GB allocated)

 

can anyone help me figure this out please?

[Mark Syms]

Aside from XenServer 7.6 being long since End of Life.

 

With 1.5TB free you should have plenty of spare space to permit the coalesce (it temporarily requires extra space whilst running in order to combine the active data blocks), which means that you shouldn't be hitting the most common reason which is that lack of space. The other very common case is that the coalesce of the final part of the snapshot tree cannot be completed as the VM is writing data to the leaf node faster than the coalesce process can merge things into the parent node. This can result in each VM taking 2x the amount of space assigned to it, the leaf node is always fully committed to the size assigned to allow the VM to write wherever it needs to and if the parent node is full of data and/or dirty data blocks then it can be the same size (or very close to).

 

There have been some improvements in this area in Citrix Hypervisor 8.2, notably a dynamic limit control on when the leaf coalesce decides to pause the VM for the final offline coalesce, this helps in a lot of cases but is not completely guaranteed to complete if the VM is very busy performing writes.

[scott budworth]

if the VM is wriring data fast, how long should i shut it down for to complete the process? - Tried shutting down the VM for a while, no avail. 

 

also attempted to process mahually: 

 xe host-call-plugin host-uuid=c57163c2-c1b0-43ab-b5ce-0c61d587d8dc plugin=coalesce-leaf fn=leaf-coalesce args:vm_uuid=76668f62-94eb-f32d-b1b1-ab48426769a1
VM has no leaf-coalesceable VDIs

[Tobias Kreidl]

The coalesce process can take up to 24 hours. Another option would be to migrate at least some of storage to other SRs, clean up that SR, and then migrate the storage back again.

 

-=Tobias

[Mark Syms]

I'm 80-90% sure it's not that there insufficient space in the SR (there is 1.5TB free). It's much more likely IMO that it's the GC not keeping up with the VM write. I can't remember what the log looks like in 7.6 when the GC gives up but there's probably something in /var/log/SMlog to give a hint.

[scott budworth]

hi guys, thank you for all your replies. 

 

i have left it all day and storage has dropped down by 30gb so it must be doing somthing or maybe its released space from a small image, only time will tell. 

 

i will drop you all an update tomorrow. thanks again!

[scott budworth]

n/a

[Mark Syms]

That's the wrong log and not the one I mentioned in my comment above, looks like audit.log?

[scott budworth]

sorry, here is smlog 

SMlog.txt

[scott budworth]

bump

[Tobias Kreidl]

Seems it cannot see this volume, at lest at one point:

 

Failed to find logical volume "VG_XenStorage-0cc9b855-d381-5532-326c-056640009b65/bd7be201-5a25-498c-a096-bfba8c94ea6b.cbtlog"

 

Is that volume active?

[Mark Syms]

Ignore any errors about cbtlog files. XenServer 7.6 had a bug where it would noisily complain about cbtlog even when CBT wasn't enabled.