If you're looking for a support matrix, don't look here. Instead, refer to the NCP compatibility requirements links below; for a given version of NCP (VMware NSX Container Plugin), it will show the compatible versions of NSX-T Data Center and TAS.
The following tables show combinations of TAS (VMware Tanzu Application Service for VMs, formerly PAS, Pivotal Application Service), NSX-T, and NSX-T Container Plug-in (NCP). A cell with a checkbox icon is supported by VMware. A cell with a caution icon is interoperable: it works, it passes our acceptance suites, but is not explicitly supported. A cell with no icon either was not tested, did not pass our suites, or is known not to work. Not all combinations are tested. For example NCP 2.3.0 is not tested against TAS 2.5, for consumers of TAS 2.5 would likely install a more recent version of NCP. Acceptance suites include the following:
src/test/acceptance directory) which test the Application Security Groups (ASGs) (firewall rules for containers to the outside world), and Network Policies (firewall rules for container-to-container networking).Disclaimer: An interoperable combination does not imply that all Cloud Foundry services will work with that combination. Sometimes we only test a portion of the features. For example, we test Isolation Segments, but only the container placement (i.e. that apps are placed on a shared or isolated Diego cell as directed) and not the routing restrictions (i.e. that the shared routers cannot reach the isolated apps and vice-versa).
| Symbol | Meaning |
|---|---|
| VMware supported | |
| Interoperable but not supported | |
| (Blank) Untested, unsupported, or known not to work |
ncp/cluster tags from the T0 router. See this KB article for more information. Symptom: TAS fails to deploy while running the Notification Errand.cf ssh to the Windows apps is not possible by default. One can modify the BOSH job's ssh-proxy property connect_to_instance_address to false, at which point cf ssh will work to Windows apps but no longer work to Linux apps. By default, NCP places the Linux apps on a special network, but leaves the Windows apps on the original network. The cf ssh command can be globally directed to one of the two networks, but not both. This limitation is currently being addressed; to monitor progress, see the associated GitHub issue and Pivotal Tracker storycf push fails and cf logs contains the error external networker up: exit status 1, this is a possible symptom that the IP Pool is exhausted. To check the pool, log into the NSX-T manager and navigate: Advanced Networking & Security → Inventory → Groups → IP Pools.