Over time, the number of calls being made from the TCA-CP to vCD increases drastically. This is exacerbated in large scale environments where there are many OrgVdcs available.
The vCD database is heavily utilized due to a large number of incoming calls and vCD performance is degraded due to the increased load on the database.
Within the TCA-CP appliance, certain inventory pollers are re-queued and cloned after the restart of the app-engine service.
There are two workarounds that will help alleviate this issue, one is done on vCD while the second is done on the TCA-CP.
Follow the procedure outlined in Limit audit_trail to improve vCloud Director DB Performance (89933) .
1. SSH into the TCA-CP appliance.
2. Switch user to root.
su -
3. Stop the app-engine service by running the following command:
systemctl stop app-engine
4. Log into MongoDB by running the following command:
mongo hybridity
5. Delete Jobs in Mongo with the following commands:
db.Job.find({"jobType":"VcdInventoryJobType","jobData.entityType":"adminOrgVdcStorageProfile"}).count()
db.Job.deleteMany({"jobType":"VcdInventoryJobType","jobData.entityType":"adminOrgVdcStorageProfile"})
6.Steps for Service Config change:
a. Change directory to /opt/vmware/deploy/zookeeper/
b. Edit file VcdInventoryService.zql to change the frequency value under statsFeeder to 86400. This value is in seconds.
c. Edit file VcStatsService.zql to change both, the frequency value (under statsFeeder) and monitorIntervalInSecondsUnder to 86400. This value is in seconds.
d. Under the same directory edit the file createZkNode.sh and change rmr to deleteall.
e. Save the file.
f. Under the same directory execute the following command:
./createZkNode.sh VcdInventoryService
7. Restart the app-engine service with the following command:
systemctl restart app-engine
Note: Do not restart the app-engine service on the TCA-CP appliance frequently as this will create the cloned jobs again.