HPE StoreVirtual Storage / LeftHand
cancel
Showing results for 
Search instead for 
Did you mean: 

StoreVirtual OS 12.7 is out

vlho
Advisor

StoreVirtual OS 12.7 is out

Hi all,

download: https://h20392.www2.hpe.com/portal/swdepot/displayProductInfo.do?productNumber=StoreVirtualSW

News: support for VMware ESX 6.5 and Windows Server HyperV hypervisors as well as integration with HPE Recovery Manager Central.

Next new features? I don't know, beacuse link for Relase Notes 12.7 point to old version 12.6. Maybe it will be OK tomorrow...

 

7 REPLIES
SVprodmgr
Advisor

Re: StoreVirtual OS 12.7 is out

The release notes for StoreVirtual OS 12.7 will be released by the end of May.  This is a minor release in which we add support for ESX 6.5.  Additionally, we have integrated StoreVirtual VSA with HPE's RMC/StoreOnce and with HPE's VM Explorer.  There are a few enhancements in the release too.

Amy Mitchell

StoreVirtual OS Product Manager

I'm an HPE employee working in product management
Mike Chivers
Occasional Advisor

Re: StoreVirtual OS 12.7 is out

I have installed v12.7 as an upgrade from 12.6 on ESXi6.5 and it all went well.

Does anyone know if I can use the ESXi6.0 version of the iSCSI rescan vib on a ESXi6.5 install?

metadata-hpe-iscsi-rescan_6.0.0-12.0.0.1-offline_bundle.zip / hpe-iscsi-rescan-mem-6.0.0-12.0.0.1.vib

vlho
Advisor
rgreen83
Occasional Visitor

Re: StoreVirtual OS 12.7 is out

I installed 12.7 (from 12.6) on our 4 node 4335 cluster with 10gbe and a 2 node 4330 cluster with 1gbe on 5/27. The 4330 cluster has been in production for several years and the 4335 cluster has been in production since February, and have had no issues with either until now.  All hosts are hpe gen8 and gen9 dl380s, 10gbe switches are 5700 direct attach to hpe 560sfp+ nics, and hpe 2920 switches for the 1gbe side. ESX 6.5 since the 4335 cluster was built.

Ever since installing 12.7 I have received emails every 5 minutes about packet loss from all 4 nodes on the 4335 cluster,like  EID_IP_ERRORS_EXCESSIVE 'Excessive' packet loss. Packet Loss is ' 3.03'%., and had a node take itself offline Monday overnight. From what I have read it was common for this to popup back when 12.6 was released when they added this event condition, but I've always been on 12.6 and never had it before 12.7

We've been working with HP storage and networking support for a week and still have no resolution. The networking guy says the 5700 10gbe switches are configured perfectly and there are zero dropped packets there. 

Just wanted to offer a word of warning for anyone looking to jump to 12.7 right away, you can't easily go back so don't rush into it like I did. If support finds a resolution I'll try to post it back here, but I have to wait another 4 days for them to make a tier 3 storage support engineer available.

Admin85858
Occasional Visitor

Re: StoreVirtual OS 12.7 is out

Same here since upgrading from 12.5 to 12.7. betweeen 1 to 3% loss.

Only appears when running Backup Job at night!

How can I manage to find out where the failure is located?

rgreen83
Occasional Visitor

Re: StoreVirtual OS 12.7 is out

Update to my previous post. I ended up getting this fixed myself. 

We did eventually hear back from a tier 3 storage support engineer, but not directly and he had only indirectly viewed our config files, after over a week not a single person at HP was willing to share a screen or anything to look at the CMC directly. I have to assume if they are the experts this should have been something they should have been able to see as an issue pretty early on if they had just looked at it. Their recommended solution was to downgrade the CNA firmware back to a 2015 firmware, I didn't end up doing that.

It turns out the packet loss notices had nothing to do with packets being dropped in the network, it is either a timing or buffering issue, something that changed between 12.6 to 12.7. As noted before I have 10gbe on the iscsi side and 1gbe on the management side, somewhat in the middle of those is the "lefthand os interface" which is where the nodes talk to each other. What I did not previously know is that the lefthand os interface is also what is using for balancing, syncing, etc. I originally had this interface for each node on the management side, having not wanted to clutter up the iscsi network if possible and also because this was required for our prior cluster that had a FOM because the FOM has no iscsi connectivity and is really only there as a manager and quorom witness.

After changing the lefthand os interface to the 10gbe bond on each node (so it is sharing the iscsi bond now) I have received zero packet loss notifications. Not sure why this was never an issue on 12.6 or if this is what HPE considers best practice since they offer little detail about the 3 interfaces and how they should be setup, but I've had no issues in the last 24 hours with this configuration. Hope this helps someone else.

SVprodmgr
Advisor

Re: StoreVirtual OS 12.7 is out

Thank you for updating about how you resolved the networking events.  The StoreVirtual 12.6 User Guide http://h20566.www2.hpe.com/hpsc/doc/public/display?sp4ts.oid=4118659&docLocale=en_US&docId=emr_na-c05049834 provides good information about the best practices for networking.  It explains the different traffic types and recommends that management traffic be separate from iSCSI and StoreVirtual/LeftHand traffic.  This User Guide applies to Release 12.7 as well as 12.6.

Amy Mitchell

HPE StoreVirtual Product Manager

I'm an HPE employee working in product management