Description of problem: Registering hypervisors with the same ID to different orgs is blocked. If two virt-who instances in different orgs report the same hypervisorId, the second one will currently fail. Version-Release number of selected component (if applicable): all versions currently (latest is 9.26.1) I have a more permanent fix that should be in candlepin master soon, but I don't think we can get it into sat6 for release. How reproducible: 100% Steps to Reproduce: point two virt-who instances (in different orgs, registered to the same sat6) at the same cluster. second one to run will always fail. Actual results: One virt-who fails to create the host-guest association. It receives errors (result.failedUpdate in the json response), which should also appear in candlepin logs (info level) Expected results: Each organization should have a hypervisor consumer created to represent the given hypervisor ID Additional info: The fix shouldn't involve any code change. If we append this line to the installed candlepin.conf, the problem should be resolved: candlepin.block_duplicate_hypervisor_ids = false Future candlepin versions should remove that config and always behave as though 'false' is used.
Since this issue was entered in Red Hat Bugzilla, the release flag has been set to ? to ensure that it is properly evaluated for this release.
But wait, there's more!! The candlepin server included in satellite 6 is missing default configuration values for c3p0. Without any configuration, too many concurrent requests will cause candlepin to request more database connections than the database can handle, causing the request to fail. # Default values for c3p0 to avoid requesting more connections than possible # Without this, problems can occur in heavy load scenarios (attempts to get a db connection will cause exceptions) jpa.config.hibernate.connection.provider_class=org.hibernate.connection.C3P0ConnectionProvider jpa.config.hibernate.c3p0.min_size=5 jpa.config.hibernate.c3p0.max_size=20 jpa.config.hibernate.c3p0.timeout=300 jpa.config.hibernate.c3p0.idle_test_period=300 jpa.config.hibernate.c3p0.max_statements=0
This bug has been verified on RHEL-7.1-20141024.1 against sat6 zstream but failed, please check following steps: 1, Configure virt-who host1 with ESX and register to Satellite, org selected: Default_Organization 2, Start virt-who service in host1, check hypervisor consumer created in Default_Organization --> Content Hosts page 3, Configure virt-who host2 with the same ESX and register to Satellite, org selected: sgao_org 4, Start virt-who service in host2, check hypervisor consumer created in sgao_org --> Content Hosts page ---- This step failed, the second hypervisor consumer did not show up here, instead in Default_Organization --> Content Hosts page, a duplicated hypervisor consumer has been added. 5, Check log files /var/log/rhsm/rhsm.log in both the 2 virt-who hosts, no error show up.
Created attachment 953014 [details] capture Please check the captures
Are the candlepin logs from the satellite available?
This limitation was removed prior to Sat-6.0 release.