Unassigned Hosts and Network Pools not visible under the Global Inventory List in the Management vCenter when logged in with VCF SSO user.
search cancel

Unassigned Hosts and Network Pools not visible under the Global Inventory List in the Management vCenter when logged in with VCF SSO user.

book

Article ID: 425591

calendar_today

Updated On:

Products

VCF Operations VMware vCenter Server VMware Cloud Foundation

Issue/Introduction

  • When logging into the VCF Management Domain vCenter UI using a VCF SSO user, the following components are not visible:

    • Unassigned Hosts: Global Inventory List -> Hosts -> Unassigned Hosts
    • Network Pools: Global Inventory List -> Hosts -> Network Pools

  • The Unassigned Hosts and Network Pools web pages return the following unauthorized error:
    Authorization Error : Unauthorized access.
    There is a restriction with the page you are trying to reach and it cannot be displayed.

     

  • The VCF SSO user or their group is part of the SDDCAdmins group in vCenter and that the user or group is assigned the Single Sign On Admin Role in SDDC Manager
  • The same behavior can be observed in vCenter password management page failing to load the component passwords and instead displaying the following error:

    Failed to load password management data. Cannot convert a Symbol value to a string Failed to load PSC data.
    Cannot convert a Symbol value to a string



  • vsphere_client_virgo.log highlights either of the below errors, where "sddcmgr.domain" is the SDDC Manager that should be looked into regarding this KB's issue.
    /var/log/vmware/vsphere-ui/logs/vsphere_client_virgo.log

    YYY-MM-DD] [WARN ] -nio-127.0.0.1-####-exec-###  com.vmware.vsphere.client.rest.SsoTokenController                 Error pushing SAML token to web hook com.vmware.vcf.client:version.and.build:########## of plugin https://sddcmgr.domain:443/plugin/api/saml-hook; thumbprint=null; certificate=.
    org.springframework.web.client.HttpClientErrorException$BadRequest: 400 Bad Request: "<html><EOL><EOL><head><title>400 Request Header Or Cookie Too Large</title></head><EOL><EOL><body><EOL><EOL><center><h1>400 Bad Request</h1></center><EOL><EOL><center>Request Header Or Cookie Too Large</center><EOL><EOL><hr><center>nginx</center><EOL><EOL></body><EOL><EOL></html><EOL><EOL>"
    
    YYYY-MM-DD7THH:MM:SS.sssZ WARNING vsphere-ui 330971 [vc@#### threadName="http-nio-127.0.0.1-####-exec-###" logger="com.vmware.vsphere.client.rest.SsoTokenController"  ] Error pushing SAML token to web hook com.vmware.vcf.client:version.and.build:######
    ### of plugin https://sddcmgr.domain:443/plugin/api/saml-hook; thumbprint=null; certificate=-org.springframework.web.client.HttpServerErrorException$InternalServerError: 500 Internal Server Error on POST request for "https://sddcmgr.domain:443/plugin/api/saml-hook": [no body]

Environment

VMware Cloud Foundation 9.0.x
VCF Operations 9.0.x
vCenter Server 9.0.x

Cause

The issue is encountered due to the SAML token length exceeding the default nginx buffer size configuration.

Resolution

The issue is under Broadcom Engineering review and the fix is expected to be included in the future release. 

Workaround:

  1. Take snapshot of the SDDC Manager. 

  2. SSH to sddc-manager using vcf user and elevate to root.

  3. Take a backup of nginx.conf:    
    cp /etc/nginx/nginx.conf /etc/nginx/nginx.conf.backup

     

  4. Add the below two lines under the 'http' section of nginx.conf:
    These lines must be added between map_hash_bucket_size and proxy_send_timeout
    client_header_buffer_size 4k;
    large_client_header_buffers 4 32k;
    • Reference:

      Original File (before any changes):

      http {
          limit_conn_zone $server_name zone=per_server:10m;
          limit_conn per_server 1000;
          limit_conn_zone $binary_remote_addr zone=per_ip:10m;
          limit_conn per_ip 100;
          limit_req_zone $binary_remote_addr zone=api_traffic:10m rate=5000r/s;
          limit_req_zone $binary_remote_addr zone=ui_traffic:10m rate=1000r/s;
          limit_req_zone $binary_remote_addr zone=ui_file_upload:10m rate=1r/s;
          map_hash_bucket_size 128;
          proxy_send_timeout 600;
          proxy_read_timeout 600;

       

    •  Updated File (after the above changes):

      http {
          limit_conn_zone $server_name zone=per_server:10m;
          limit_conn per_server 1000;
          limit_conn_zone $binary_remote_addr zone=per_ip:10m;
          limit_conn per_ip 100;
          limit_req_zone $binary_remote_addr zone=api_traffic:10m rate=5000r/s;
          limit_req_zone $binary_remote_addr zone=ui_traffic:10m rate=1000r/s;
          limit_req_zone $binary_remote_addr zone=ui_file_upload:10m rate=1r/s;
          map_hash_bucket_size 128;
          client_header_buffer_size 4k;
          large_client_header_buffers 4 32k;
          proxy_send_timeout 600;
          proxy_read_timeout 600;

       

  5. Restart nginx with the below command to pick up the above changes:
    nginx -s reload 

     

  6. Use an incognito or private window to confirm that Global Inventory Lists populates now as a VCF SSO user.
    • Note: You might need clear your browser cache and cookies to ensure the vSphere Client reflects the changes

 

If the issue persists:

  1. Ensure that the SSO user or group has the permissions noted in the Issue Summary for SDDCAdmins and Single Sign On in SDDC Manager.
  2. If you are using linked vCenters with multiple SDDC managers, ensure that all SDDC managers have this nginx.conf change.
  3. Confirm that nginx was restarted after making the above changes in nginx.conf for each SDDC manager as applicable.