Unsolved
This post is more than 5 years old
2 Intern
•
359 Posts
1
1395
VNX5200 non disruptive software upgrade?
I have a VNX 5200 redundant with dual control stations, etc... serving 60+ VMWare virtual machines across 6 hosts using 10gbe NFS via filesystems NFS export.
Currently running on 8.1.2-51 for file and 5.33.000.5.052 for block.
I see that release 8.1.6.101 for file and 5.33.006.5.102 for block are out.
Would Unisphere Service manager have the ability to download and safely upgrade a working VNX5200 without any down time, or would this have to be scheduled for after hours, and if so how long of a maintenance window would one need? To power down 60 VM's is a tough proposition to get scheduled. Downtime would need to be minimized.
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
June 29th, 2015 12:00
how many file systems/snapshots on your datamovers ? Failover times on datamovers are pretty fast, less than a minute so NFS datastores will hang while file system is remounted on standby datamover. VMware is pretty good at handling these failovers. I would still perform all upgrades during a scheduled maintenance window.
kparrotte
40 Posts
1
June 29th, 2015 12:00
Hi
You should also pay pretty strict attention to the VMware Integration Guide in terms of key settings to help make VMware less vulnerable to potential disruptions based upon incorrect time out settings; which we see a lot.
It is a pretty large pdf file but mainly because there are a lot of screen shots inside: https://support.emc.com/docu33753_Using-VNX-Storage-with-VMWare-vSphere.pdf?language=en_US
Ought to be required reading.
Best wishes.
kjstech
2 Intern
2 Intern
•
359 Posts
0
June 29th, 2015 13:00
Yes we have the registry edit to change the windows scsi timeout to 120.
Last upgrade I did was just a block only upgrade, but none of the machines even batted an eye. I did it on a Saturday afternoon when no one was around just to make sure, but a constant ping and monitoring of VM's concluded there was no affect. Though again that was just upgrading block code, there wasn't an update to file code at the time.
Thank you for the document again. Since that time we've rolled out quite a few Windows Server 2012 R2 VM's and I'm not 100% sure that the scsi timeout was changed to 120 in that VM template those machines derived from. I'll do a quick check.
We have 3 NFS exported filesystems, one on each of their own subnet.
They are 4TB in size with autogrowth at 90% full to a max of 8TB.
We are not running CIFS on this box at this time.
We do not use snapshots on VNX5200 platform at this time. We do use VMWare snapshots but they are always pruned and never kept for very long (that can get very messy).
For replication we are using VMWare replicator as it is storage hardware agnostic and replicates to an older NX4 Celerra offsite.
NFS transport is on a jumbo frame 10Gbe isolated Ethernet network (Brocade Turbo Iron switch).