Simple operations such as Power on/off VMs, Import VM from vCenter are taking much longer time than the average and this is impacting the performance of the vCD.
Cloud Director 10.3.x/10.4.x is impacted by an ongoing issue where the Artemis broker, that allows the communication between nodes fails and slowness starts as other cells rely on polling vCenter to receive updates about the task progress.
Engineering has added the fix in 10.5.0 Cloud Director versions GA now from Customer portal.
The workaround available is to modify the following parameters from the primary cell:
- SSH to the Primary cell using Putty or other tool.
- Set the “vc-task-completions-retrieval-timer-interval-sec” parameter to 10.
cell-management-tool manage-config -n vc-task-completions-retrieval-timer-interval-sec -v 10 - Set the “vcloud.activities.activityRelayPollingIntervalMs” paramter to 10000
cell-management-tool manage-config -n vcloud.activities.activityRelayPollingIntervalMs -v 10000 - Restart all Cloud Director cells so change can take effect.
Note: Appliance’s cluster failover recommended to be in “Manual” mode to restart the cells.
