Adding VxRail host/Cluster expansion to SDDC Manager takes an extended amount of time while fetching portgroups information for hosts from vCenter.

Products

• VMware SDDC Manager

Issue/Introduction

Adding the host is taking a long time as we are fetching portgroups information for hosts from vCenter by prefix and For vSAN and vMotion transport type, portgroup names do not match exactly, due to which we traverse over all portgroups for each host in vCenter resulting in the workflow taking a huge amount of time (resulting in hours/days based on the number of portgroups).

In Domain Manager logs, We can observe a similar log as shown below:

 INFO  [vcf_dm,6808f6927e9a29749192c6b73fa04f17,2792] [c.v.v.v.p.service.HostSystemPlugin,dm-exec-11]  host name <sample_host_name.domain.local>
 INFO  [vcf_dm,6808f6927e9a29749192c6b73fa04f17,2792] [c.v.v.v.p.service.HostSystemPlugin,dm-exec-11]  value of nic size for host <sample_host_name.domain.local>, 4, [NicInfo(name=management.key-vim.host.VirtualNic-vmk2, type=management, ipAddress=XX.XX.XX.XX, defaultGateway=XX.XX.XX.XX, subnetMask=XX.XX.XX.XX, portGroup=Management Network-XXXX, portGroupKey=dvportgroup-XX), NicInfo(name=management.key-vim.host.VirtualNic-vmk0, type=management, ipAddress=, defaultGateway=XX.XX.XX.XX, subnetMask=, portGroup=VxRail Management-XX.XX.XX.XX, portGroupKey=dvportgroup-XX), NicInfo(name=vmotion.key-vim.host.VirtualNic-vmk4, type=vmotion, ipAddress=XX.XX.XX.XX, defaultGateway=XX.XX.XX.XX, subnetMask=XX.XX.XX.XX, portGroup=null, portGroupKey=null), NicInfo(name=vsan.key-vim.host.VirtualNic-vmk3, type=vsan, ipAddress=XX.XX.XX.XX, defaultGateway=XX.XX.XX.XX, subnetMask=XX.XX.XX.XX, portGroup=null, portGroupKey=null)]

The log message indicates that the portGroup is null for both vSAN and vMotion types. As a result, the system traverse through all portgroups, which is causing the delay.

Environment

VCF on VxRail 5.X

Cause

Add host is fetching portgroups information for hosts by prefix and the reason that the prefix does not match is customer has modified portgroup names for vSAN and vMotion type after adding the host to VxRail.

 

Resolution

Currently, the code fetches host portgroups information from vCenter using a prefix. The workaround below will modify it to fetch portgroups by MoR instead of by-prefix, and then test the workflow.

Workaround Steps :

1. Take SDDC Manager VM snapshot.
2. SSH into the SDDC Manager VM and get specified features by using the command below:
   curl http://localhost:7200/domainmanager/features/list | json_pp | grep -E 'feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy|feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy'

output:
   "feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy" : "false",
"feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy" : “true”,
   
   
3. Navigate to /home/vcf directory and execute command:
   vi feature.properties
   
   
4. Add the following 2 lines and save the file:
   feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy=true
feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy=false
   
   
5. Ensure that the owner and group permissions for feature.properties are set to vcf:vcf, as shown below:
   
   -rw-r--r-- 1 vcf  vcf  203 May  5 00:00 feature.properties


6. Restart commonsvcs, operations manager and domain manager services in the specified order.
   
   
7. Execute the following command to retrieve the features and verify that ""feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy"" is set to true and ""feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy"" is set to false.
   
   curl http://localhost:7200/domainmanager/features/list | json_pp | grep -E 'feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy|feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy'

Output
   "feature.vxrail.cluster.discovery.optimisation.queryPGByMoRStrategy" : "true",
   "feature.vxrail.cluster.discovery.optimisation.queryPGByPrefixStrategy" : "false",
   
   
8. Start new add host/ cluster workflows.