ON THIS PAGE
Known Behavior
This section contains the known behavior and limitations for the specified Contrail Cloud Release.
Known Behaviors for Contrail Cloud Release 13.5
This section contains known behaviors and limitations in Contrail Cloud Release 13.5.
The following capabilities available in Contrail Networking Release 1912.L3 are not supported in Contrail Cloud 13.5:
Contrail Fabric Management. For more information on fabric management, see:
Contrail Enterprise Multicloud support for AWS and Azure.
Contrail Command deployment - only the UI is supported for Contrail Cloud Release 13.5.
Netronome SmartNIC limitations:
CC-648 - Compute node network configuration is reset after reboot.
CC-673 - Contrail status does not show Netronome Agilio container status.
CC-674 - Restart of nova_compute breaks Netronome Agilio container.
CC-675 - Hanging vif interfaces
CC-678 - VM instances stuck in BUILD state.
CC-680 - Connectivity loss after container restarts
CC-697 - SSL for undercloud integration is not working with IDM supplied certificates.
CC-589 - Using spine/leaf topology with IDM fails to generate leaf certificates.
libvirt certificates are not generated for the leaf internalAPI networks, causing deployment failures.
Certificates must be manually pre-allocated in IDM for every resource on leaf internalAPI networks.
Limited Barbican feature support in Contrail Cloud Release 13.5.
The Barbican features not included in Release 13.5 are as follows:
Horizon plugin for Barbican.
Barbican object storage encryption.
Known Behaviors for Contrail Cloud Release 13.4
This section contains known behaviors and limitations in Contrail Cloud Release 13.4.
The following capabilities available in Contrail Networking Release 1912.L2 are not supported in Contrail Cloud 13.4:
Contrail Fabric Management. For more information on fabric management, see:
Contrail Enterprise Multicloud support for AWS and Azure.
Using Contrail Command for Contrail Cloud deployment.
Command UI is supported for Contrail Cloud 13.4 Release, but the deployment capability is not compatible with Contrail Cloud deployment.
Contrail Cloud 13.4 is released with a Red Hat content sync of 18 May 2020. Resolutions for the following security advisories are not yet available with Contrail Cloud 13.4:
CVE-2020-8617 Red Hat CVSS v3 Score - 7.5
CC-473 - Live migration of VMs is broken and is a known issue in OpenStack Queens. The error was reproduced in a lab environment and was tested with kernel compute to kernel compute, and DPDK compute to DPDK compute. The lab testing produced the error shown below:
(overcloud) [stack@undercloud ~]$ openstack server migrate --live overcloud8wq-compkernel1hw0-1.cc13-testbed-1.juniper.net kmod0-vm1Migration pre-check error: Instance has an associated NUMA topology. Instance NUMA topologies, including related attributes such as CPU pinning, huge page and emulator thread pinning information, are not currently recalculated on live migration. See bug #1289064 for more information. (HTTP 400) (Request-ID: req-27d8710a-7901-41c9-a801-c0f0b6acc5cf)
For more information, see: Red Hat Bugzilla - Bug 1196242, and Launchpad Nova Bug 1289064.
CC-509 - Open vSwitch (OvS) update on control host nodes causes the network connectivity to break. This issue happens during post-deployment activities of the manual yum update and reboot of the control host.
A workaround for this issue is to temporarily switch the bridge used for the control plane/provisioning network to Linux bridge. See the procedure below for the workaround.
On the Control Host node:
Create a backup copy of the devices used in the Control/Provisioning communication: both bridge and bridge interface.
Apply the following changes:
For the bridged interface you will comment out DEVICETYPE, TYPE, and OVS_BRIDGE. Then you will add TYPE=Ethernet.
# DEVICETYPE=ovs # TYPE=OVSPort # OVS_BRIDGE=br-ens2f0 TYPE=Ethernet
For the bridge comment out DEVICETYPE and TYPE. Then you will add TYPE=Bridge.
# DEVICETYPE=ovs # TYPE=OVSBridge TYPE=Bridge
Use tmux, screen, or nohup to separate command execution from ssh connection status.
The commands below will use nohup to remove OvS bridge and apply the new Linux bridge configuration that was completed in the previous step:
nohup sh -c 'ovs-vsctl del-port <BRIDGE_NAME> <INTERFACE>; ovs-vsctl del-br <BRIDGE_NAME>; ifup <BRIDGE_NAME> sudo ifup <INTERFACE>' &
Check interfaces status (interfaces should be visible as per Linux bridge configuration).
Perform the yum update.
Restore the backup that were prepared in the first step.
Perform a reboot to reload the kernel and OvS configuration.
Known Behaviors for Contrail Cloud Release 13.3
This section contains known behaviors and limitations in Contrail Cloud Release 13.3.
The following capabilities available in Contrail Networking Release 1912.L1 are not supported in Contrail Cloud 13.3:
Contrail Fabric Management.
Public cloud deployment and connectivity.
Using Contrail Command for Contrail Cloud deployment.
Command UI is supported for Contrail Cloud 13.3 Release, but the deployment capability is not compatible with Contrail Cloud deployment.
TLS certificate management using IDM for Contrail Services.
Contrail Cloud 13.3 is released with a Red Hat content sync of 25 Feb 2020. Resolutions for the following security advisories are not yet available with Contrail Cloud 13.3:
CVE-2018-1000076 Red Hat CVSS v3 Score - 5.5
CVE-2018-1000078 Red Hat CVSS v3 Score - 6.1
CVE-2019-14818 Red Hat CVSS v3 Score - 7.5
CVE-2019-15605 Red Hat CVSS v3 Score - 7.1
CVE-2020-8597 Red Hat CVSS v3 Score - 9.8
CC-473 - Live migration of VMs is broken and is a known issue in OpenStack Queens. The error was reproduced in a lab environment and was tested with kernel compute to kernel compute, and DPDK compute to DPDK compute. The lab testing produced the error shown below:
(overcloud) [stack@undercloud ~]$ openstack server migrate --live overcloud8wq-compkernel1hw0-1.cc13-testbed-1.juniper.net kmod0-vm1Migration pre-check error: Instance has an associated NUMA topology. Instance NUMA topologies, including related attributes such as CPU pinning, huge page and emulator thread pinning information, are not currently recalculated on live migration. See bug #1289064 for more information. (HTTP 400) (Request-ID: req-27d8710a-7901-41c9-a801-c0f0b6acc5cf)
For more information, see: Red Hat Bugzilla - Bug 1196242, and Launchpad Nova Bug 1289064.
CC-509 - Open vSwitch (OvS) update on control host nodes causes the network connectivity to break. This issue happens during post-deployment activities of the manual yum update and reboot of the control host.
A workaround for this issue is to temporarily switch the bridge used for the control plane/provisioning network to Linux bridge. See the procedure below for the workaround.
On the Control Host node:
Create a backup copy of the devices used in the Control/Provisioning communication: both bridge and bridge interface.
Apply the following changes:
For the bridged interface you will comment out DEVICETYPE, TYPE, and OVS_BRIDGE. Then you will add TYPE=Ethernet.
# DEVICETYPE=ovs # TYPE=OVSPort # OVS_BRIDGE=br-ens2f0 TYPE=Ethernet
For the bridge comment out DEVICETYPE and TYPE. Then you will add TYPE=Bridge.
# DEVICETYPE=ovs # TYPE=OVSBridge TYPE=Bridge
Use tmux, screen, or nohup to separate command execution from ssh connection status.
The commands below will use nohup to remove OvS bridge and apply the new Linux bridge configuration that was completed in the previous step:
nohup sh -c 'ovs-vsctl del-port <BRIDGE_NAME> <INTERFACE>; ovs-vsctl del-br <BRIDGE_NAME>; ifup <BRIDGE_NAME> sudo ifup <INTERFACE>' &
Check interfaces status (interfaces should be visible as per Linux bridge configuration).
Perform the yum update.
Restore the backup that were prepared in the first step.
Perform a reboot to reload the kernel and OvS configuration.
CC-514 - This issue is seen when updating a non-HA topology from CC13.2 or CC13.2.1 to CC13.3. Previous releases assumed fencing was enabled even when HA was not part of the topology. See below for an example of the error output:
TASK [overcloud : Run prepare-update.sh] *************************************** Thursday 30 April 2020 20:18:15 -0700 (0:00:03.928) 0:05:10.076 ******** fatal: [192.168.122.39]: FAILED! => {"changed": true, "cmd": "source /home/stack/stackrc && /home/stack/prepare-update.sh", "delta": "0:00:03.140025", "end": "2020-04-30 20:18:19.081609", "failed_when_result": true, "msg": "non-zero return code", "rc": 1, "start": ...... "Error: The following files were not found: /home/stack/tripleo-heat-templates/environments/fencing.yaml"]}
Use the CLI input below as a workaround for non-HA updates from CC13.2/CC13.2.1 to CC13.3:
sed -i '/^fencing/d' /var/lib/contrail_cloud/facts.d/overcloud.fact
Known Behaviors for Contrail Cloud Release 13.2
This section contains known behaviors and limitations in Contrail Cloud Release 13.2.
The following capabilities available in Contrail Networking Release 1910 are not supported in Contrail Cloud 13.2:
Contrail Fabric Management.
Public Cloud.
Contrail Command Deployment–only the UI is supported for Contrail Cloud 13.2 release.
TLS certificate management using IDM for Contrail Services.
Contrail Cloud 13.2 is released with a Red Hat content sync of 1 Oct 2019. Resolutions for the following security advisories are not yet available with Contrail Cloud 13.2:
CVE-2018-1102 Red Hat CVSS v3 Score - 9.9
CVE-2018-1002105 Red Hat CVSS v3 Score - 9.8
CVE-2019-6778 Red Hat CVSS v3 Score - 7.8
CVE-2018-14632 Red Hat CVSS v3 Score - 7.7
CVE-2018-20815 Red Hat CVSS v3 Score - 7.0
CVE-2018-1127 Red Hat CVSS v3 Score - 4.2
The deployment of Contrail Command requires connectivity to an external package repository outside of the Contrail Cloud Satellite.
Setting Linux bridges with DHCP causes race conditions with the underlying bridge interfaces.
Ceph Pools PGs and replica sets are not updated. After Ceph changes (new OSDs or change in the replicas) TripleO will not make adjustments. These changes must be manually applied post-deployment.
ceph-osd fsid mismatch on redeployed environment due to use of legacy filestore.
Cleaning done by Ironic during openstack-deploy-cleanup does not bring osd disk to a usable state.
Legacy filestore should not be used. BlueStore with LVM is the recommended approach.
For more information, see: Ceph Issue #22354 and Red Hat Bugzilla #1590526.
Exception is raised when rabbitmq-client tries to delete a queue for which parameters have changed. The following might display when you run
contrail-statuson the Control nodes:control: initializing (Number of connections:2, Expected:3 Missing: Database:Cassandra, IFMap Server End-Of-RIB not computed, No BGP configuration for self)
Identify the stuck queues by locating which one <queue_name> has a large number of messages. To generate the list:
sudo docker exec -it contrail_config_rabbitmq rabbitmqctl list_queues | egrep "control|collector"
Purge each problematic queue by running:
sudo docker exec -it contrail_config_rabbitmq rabbitmqctl purge_queue <queue_name>
Verify the problem has been fixed by running
contrail-statuson the Control node.It might be necessary to restart the RabbitMQ container to fix the issue. To restart the RabbitMQ container:
docker restart contrail_config_rabbitmq
Known Behaviors for Contrail Cloud Release 13.1
This section contains known behaviors and limitations in Contrail Cloud Release 13.1.
The following capabilities available in Contrail Networking Release 1908 are not supported in Contrail Cloud 13.1:
Contrail Fabric Management.
Public Cloud.
Contrail Command Deployment–only the UI is supported for Contrail Cloud 13.1 release.
TLS certificate management using IDM for Contrail Services.
Contrail Cloud 13.1 is released with a Red Hat content sync of 5-August-2019. The following security advisories were introduced after this date and not included in Contrail Cloud 13.1:
Linux bridges can only be configured with static IP.
ceph-osd fsid mismatch on redeployed environment.
Cleaning done by Ironic during
openstack-deploy-cleanupdoes not bring osd disk to a usable state.BlueStore with LVM is the preferred approach.
For more information, see: Ceph Issue #22354 and Red Hat Bugzilla #1590526.
Ceph Pools PGs and replica sets are not updated. After Ceph changes (new OSDs or change in the replicas) TripleO will not make adjustments. These changes must be manually applied post-deployment.
Running the overcloud-validation.sh script will create objects that are not subsequently cleaned up after the test completes. You must manually remove any remaining objects after the overcloud-validation script completes.
Known Behaviors for Contrail Cloud Release 13.0.2
This section contains known behaviors and limitations in Contrail Cloud Release 13.0.2.
Only the first element of the
global.ntplist inconfig/site.ymlis used when setting up NTP synchronization in the infrastructure. This entry must be reachable and a fully functional NTP server. The full list is utilized in the overcloud.The
samples/site.ymlincorrectly shows theovercloud.registry. This should be commented in the deploymentconfig/site.yml.
Known Behaviors for Contrail Cloud Release 13.0
This section contains known behaviors and limitations in Contrail Cloud Release 13.0.
Object storage for Ceph—Rados Gateway is not functional when ceph.support is set to enabled.
Contrail Cloud Release 13 configures only endpoint TLS encryption but not internal Contrail TLS.