Примечания к выпуску для Cisco Cisco Elastic Services Controller 1.0

D R A F T — C I S C O C O N F I D E N T I A L

Cisco Elastic Services Controller 1.0 Release Notes

Limitations

•

Keepalived is switching its state to BACKUP but our internal state still maintains the current

status to MASTER, the state transition triggers another reboot

Feature Limitations

•

General:

–

If SERVICE_UPDATE config fails due to OpenStack internal errors, the ESC database and the
CDB database will become out of synchronization.

–

ESC 1.0 will not accept requests to add or remove VM from an existing deployment.

•

ESC In-Flight Transactions:

–

Upon an ESC shutdown, reboot or failover, all in-flight northbound requests will be lost or
suspended. The northbound client needs to retry deployment request with ESC or request to
rollback. The northbound client can simply leverage request ‘timeout’ (or) ‘retry’ to handle this
scenario.

•

VIM Error Handling:

–

VIM related operations, will be retried based on specified configuration parameters. If the VIM
connections are not re-established during that period, ESC will proceed with its operations,
which may result in discrepancies between ESC resources view and the actual VIM resources
view. This resource leak will be reported by ESC.

•

VIM Resources Management:

–

Resources managed by ESC, should not be manipulated by out of band processes or operations.

•

VIM and ESC Resources Leakage:

–

See

•REST API:

section: Shared Network (or) Subnet Creation.

•

VNF Life Cycle Management:

–

If a service deployment fails, ESC will keep its information in its database (with error state) until it
receives an undeployment request. The undeployment will clean (or) remove objects that are in error
states.

–

If VMs on OpenStack are deleted even before a ‘VM_ALIVE’ is received from them, it is to be
noted that the some of the VM fails to recover.

–

If the interfaces associated with the VM are down or if the VM is created in an ERROR state,
the VM_Recovery is not triggered.

–

Any scale operations (scale in or scale out), may affect the normal functionalities of VM or VNF
operations (start, stop, reboot, enable, monitor, disable, monitor).

•

VM Affinity and Anti-Affinity:

ESC 1.0, supports inter-group affinity, anti-affinity, intra-group affinity and anti-affinity.

–

For intra-group, a host as the target compute is supported for the deployment.

–

For anti-affinity, there is no support for specifying the host.

–

For affinity, with host specified, deployments will result in a successful state on the target
compute host.

Note

Specifying a compute host in this way limits ESC's ability to recover the VM if the specified host becomes
unavailable.