As of March 29th, the long anticipated NSX 6.3.6 release is now available to download from VMware. NSX 6.3.6 with build number 8085122 is a maintenance release and includes a total of 20 documented bug fixes. You can find details on these in the Resolved Issues section of the NSX 6.3.6 release notes.
Aside from bug fixes, there are a couple of interesting changes to note. From the release notes:
“If you have more than one vSphere Distributed Switch, and if VXLAN is configured on one of them, you must connect any Distributed Logical Router interfaces to port groups on that vSphere Distributed Switch. Starting in NSX 6.3.6, this configuration is enforced in the UI and API. In earlier releases, you were not prevented from creating an invalid configuration.”
Since confusion with multiple DVS switches is something I’ve run into with customers in the past, I’m happy to see that this is now being enforced.
Another great addition is an automatic backup function included in 6.3.6. From the public documentation:
“When you upgrade NSX Manager to NSX 6.3.6, a backup is taken and saved locally as part of the upgrade process. You must contact VMware customer support to restore this backup. This automatic backup is intended as a failsafe in case the regular backup fails.”
As part of the upgrade process, a backup file is saved to the local filesystem of the NSX Manager as an extra bit of insurance. It’s important to note, however, that this does not remove the need to backup prior to upgrading. Consider this the backup of last resort in case something goes horribly wrong.
Another point to note is that NSX 6.3.6 continues to be incompatible with upgrades from 6.2.2, 6.2.1 or 6.2.0. You can see VMware KB 51624 for more information, but don’t try it – it won’t work and you’ll be forced to restore from backup. Upgrading to 6.2.9 before going to 6.3.6 is the correct workaround. I covered more about this issue here in a recent post.
There are a number of great bug fixes included in 6.3.6 – far too many for me to cover here, but a couple that I’m really happy to see include:
“Fixed Issue 2035026: Network outage of ~40-50 seconds seen on Edge Upgrade. During Edge upgrade, there is an outage of approximately 40-50 seconds. Fixed in 6.3.6”
This one is self-explanatory – not the expected amount of downtime to experience during an edge upgrade, so glad to see it’s been resolved.
“Fixed Issue 2058636: After upgrading to 6.3.5, the routing loop between DLR and ESG’s causes connectivity issues in certain BGP configurations. A routing loop is causing a connectivity issue. Fixed in 6.3.6”
I hope to write a separate post on this one, but in short, some loop prevention code was removed in 6.3.5, and because the AS PATH is stripped with private BGP autonomous systems, this can lead to loops. If you are running iBGP between your DLR and ESGs, this isn’t a problem, but if your AS numbers differ between DLR and ESG, you could run into this. In 6.4.0 a toggle switch was included to avoid stripping the AS PATH, so this is more of an issue in 6.3.5.