Determine whether Services vpxd depends on are not starting
Follow
How to check vCenter Server essential service status and dependencies to determine the service which is failing to start.
vCenter Server 7.0 vpxd service dependencies:
As per the diagram, in order for
vmware-vpxd to start:
- rhttpproxy and vmware-vpostgres need to start.
- If not started, then lookupsvc cannot start.
- vpxd-svcs and vmware-vpxd also depend on vmware-vpostgres.
- lookupsvc needs to start
- If not started, then vpxd-svcs cannot start.
- vmware-vpxd also depends on lookupsvc.
- vpxd-svcs needs to start.
If any of these services are failing to start first, please search for troubleshooting articles related to them. The general article for these problems is at
"503 Service Unavailable" error on the vSphere Web Client when logging in or accessing the vCenter Server.
If
vmware-vpxd is the only service failing out of this list then keep proceeding to the next section for further advice.
Determine whether disk space is full
Determine whether the problem is due to database problems
- Look for any entries containing "error vpxd" and "SQL".
- In Appliance, run this command to list the vmware-vpxd service logs entries related to SQL errors:
cat /var/log/vmware/vpxd/vpxd.log |grep "error vpxd" |grep SQL |less
- In Windows versions, this is located in: C:\ProgramData\VMware\vCenter Server\logs\vmware-vpx
- Further, look at the PostgreSQL logs.
- In Appliance versions, this is located in: /var/log/vmware/vpostgres.
- In Windows versions, this is located in: C:\ProgramData\VMware\vCenter Server\logs\vpostgres.
The PostgreSQL logs are written based on the day of the month.
Example: If today is the 15th of the month, you would look for today's entries in the file named postgresql-15.log.
- Troubleshoot the issue based on error or message:
Operating System problems
System performance problems/System resource problems
-
Detecting high storage communication latency:
When there is very average high storage communication command latency the vpxd service can stop running. The service might be powered on again but fail again at random intervals.
- Find out which ESXi host is managing the vCenter virtual machine.
- Connect as the root user to this ESXi host command line by console or via SSH client
- Run the command to check for very high latency:
cat /var/log/vmkernel.log |grep "I/O latency increased" |awk '$20 >= 80000' |less
If you see very high latency, see the " performance has deteriorated" message in ESXi for more information.
If your vCenter Server has snapshot virtual disks running on it, if there are a large number of them, or if they have grown large, it can also cause a performance impact which will cause vpxd to stop running.
Snapshots might exist which are don't show up in the vCenter or ESXi GUI
- Directly examine the files in the vCenter Server directory to ensure there aren't snapshots on the VM
Snapshot files will generally look like vmname-00000x.vmdk, where vmname is the name of the vCenter VM, and the x is the number of the snapshot.
Example: myvcenter-000001.vmdk
It is highly recommended to never keep a snapshot running on a VM for more than 3 days.
- Consolidate the vCenter VM.
-
Detecting other performance or resource problems:
- In Appliance: You can run the command vimtop to see the performance statistics of the VM.
- In Windows: You can see performance under Task Manager > Performance.
For either Appliance or Windows vCenter Servers, you can look at the virtual machines' performance history for CPU. storage, memory, and networking in the ESXi or vCenter host clients.
If there are any errors regarding resources in logs, or if the performance monitors show resources being pegged, this is suspicious for a performance or resource problem.