Xenserver 8.0 iSCSI and NFS performance issues on AOS 5.10.4

  • 4 July 2019
  • 7 replies

Good day,

Please can someone help me.
We have 2 server farms one Xenserver Pool and Nutanix Cluster
The Xenservers get its vhd storage (NFS and iSCSI) from Nutanix

Xenserver 8.0 (aka Citrix Hypervisor 8.0)
Nutanix is on AOS 5.10.4

We upgrade AOS from 5.5.4 to 5.10.4 no problems about a month later we upgrade Xenserver from 7.0 to 8.0 no problems a few days later the VMs on Xenserver started to slow down to a crawl, mean it takes 30 - 40 minutes for the VM to boot. Sometimes it just hangs and we have to force a restart. If the VM eventually come up we see in the windows resource manager 100% disk active time with less than 300KB/s throughput and Disk Queue Length between 5 -30.
Windows and Linux VMs have the same problem. Xentools drivers are up-to-date with the install files presented with Xenserver 8.0 install iso.

Before the upgrades everything was working fine. Currently we have another duplicate farm with AOS 5.10.4 but the Xenserver is still on 7.0 and that is working no issues

Does anyone have an idea what is going on?

7 replies

Userlevel 3
Badge +5
Hi @QGB007

Did you validated the compatibility between AOS 5.10.4 and XenServer 8.0? Checking the compatibility matrix, the latest XenServer version compatible is XenServer 7.5, see:

Maybe the version is not the problem, but not sure how the support can help if the running version is not supported.
Hi Richardson,

No did not validate as there is no need I think. iSCSI and NFS are standard protocols.
Xenserver is a stand-alone pool. Only the storage repository is coming from ABS
I moved to XenServer 7.1.2 LTSR same thing is happening there.
Also I don't find anything for XenServer regarding storage compatibly
Userlevel 3
Badge +5
Hi @QGB007

Sorry to disagree with you but validate compatibility of the hypervisor upgrade with the underlying storage is the most important thing to consider. I know that iSCSI and NFS are standard protocols but each storage vendor use a different set of features and this is why vendors publish HCL where they show the hypervisor versions that was tested, validated and qualified.

The following picture shows the supported XenServer versions to run with AOS 5.10.4:

Also, I don't know details about your implementation, but present Nutanix Volumes (formerly ABS) directly to XenServer hosts is not s supported configuration, here are the list of supported OS:

Supported Client Operating Systems

  • Microsoft Windows Server 2016
  • Microsoft Windows Server 2012 R2
  • Microsoft Windows Server 2008 R2
  • Red Hat Enterprise Linux 6.7
  • Red Hat Enterprise Linux 6.8
  • Red Hat Enterprise Linux 6.9
  • Red Hat Enterprise Linux 7.2
  • Red Hat Enterprise Linux 7.3
  • Red Hat Enterprise Linux 7.4
  • Oracle Linux 6.x
  • Oracle Linux 7.x
  • CentOS 7.x
  • Oracle Solaris 11.3 on SPARC
  • SuSE Linux Enterprise Server 11 / 12 (x86 servers)
  • IBM AIX 7.1 / 7.2 on POWER1


Again, I am not saying the XenServer 8.0 is the culprit here and I really recommend you open a support ticket (attaching a full NCC health check).
You provide the same type of storage to VMware but I dont see that in your list of operating systems. Does that mean its also not supported?
Userlevel 3
Badge +5
Hi @QGB007

The supported way to present Nutanix storage to VMware vSphere ESXi is mapping a Nutanix container through NFS. Mount a Nutanix container using iSCSI (Through the ABS) to VMware is not supported.

Keep in mind that mount a container to a hypervisor is different than mount Nutanix Volumes (formerly ABS) and the list of OS that supports access the Nutanix Volumes is very restricted and you can found on the link I sent in my previous post.

Again, we do not support present Nutanix Volumes to ESXi or even to Hyper-V, in case of Hyper-V we present a container through a SMB share.
Mounted container on Xenserver also slowed down to crawl