Skip to main content
Question

Prism Central Upstream Connect Error After CMSP

  • July 9, 2026
  • 4 replies
  • 194 views
W

After performing the procedure described in KB-15880 (HOW TO: Renew expired "APLOS" and "PRISMSERVICE" certificates where the last Prism Central upgrade is over two years old), we enabled CMSP (Microservices Infrastructure).

After enabling CMSP, we were no longer able to access Prism Central.

The following error message is displayed when accessing the Prism Central web console:

"upstream connect error or disconnect/reset before headers. reset reason: connection failure"

SSH access to the Prism Central VM is available, and there are no issues with the cluster status. All services appear to be running normally.

4 replies

W
  • watase
  • Author
  • Voyager
  • July 9, 2026

Could you please help identify the root cause of this issue and provide a resolution so that we can access Prism Central again?😭

BartDonders
Forum|alt.badge.img+3
  • BartDonders
  • Outrider
  • July 10, 2026

Quickest way is to contact Nutanix support…

I don’t think the community is able to help you with this limited information.

From what I read all cluster services are up and running? But what about the containers needed for CMSP?

    J
    Forum|alt.badge.img+3

    Hi,
    if this is your production I would recommend what ​@BartDonders recommended it .open a case with support let them handle it.
    but in case it is not or you dont have support.
    based on what you provided since you enabled CMSP,I would say check the mspctl health in the prism central
    login to PC as ssh  and run 
    mspctl cluster health prism-central    (do you see any component which are unhealthy?
    have a look at below article 
    A post-procedure of reconfiguring the IP address of C-MSP enabled Prism Central VMs
     

      Allan Solomon Mejia
      Forum|alt.badge.img+1

      Hello guys,

      Since this occurred immediately after enabling CMSP, I'd focus on validating the CMSP microservices rather than the traditional Prism services. Even if cluster status shows everything as healthy, one or more Kubernetes-based CMSP components may not have started correctly.

      A few things I'd check:

      • Verify the CMSP cluster health: mspctl cluster health prism-central
      • Check whether any pods are in CrashLoopBackOff, Pending, or ImagePullBackOff states.
      • Review the Envoy/Ingress and CMSP logs for connection failures, since the "upstream connect error or disconnect/reset before headers" message is often returned when the ingress proxy cannot reach a healthy backend service.
      • Confirm DNS resolution, NTP synchronization, and certificate validity, as CMSP is sensitive to these dependencies during initialization.

      If this is a production Prism Central, I'd also recommend opening a Nutanix Support case. Since the issue appeared immediately after following KB-15880 and enabling CMSP, Support can review the deployment logs and identify whether a CMSP component failed during initialization or if there's a known issue with that upgrade path.

      Terms & ConditionsAccessibility statement