4.4. RHBA-2019:0055 — Red Hat OpenStack Platform 10 Bug Fix Update

openstack-tripleo-heat-templates

BZ#1625166

This update fixes an OSP 9 to OSP 10 upgrade issue that sometimes prevented the spawning of VMs during upgrades.  

Prior to this update, VMs could not be spawned between ceph/compute upgrade and convergence, because ceph librados libraries were open in memory, conflicting with the upgraded client on disk.  That triggered calls to non-existent (in-memory) symbols.

To work around this issue, nova-compute is restarted on compute nodes to synchronize the disk and the in-memory client libraries.

BZ#1646332

This update fixes an issue that caused OpenStack API outages and control plane loss during execution of the "pcs cluster stop" command, greatly reducing the incidence of failed requests during minor updates.

Note: In manual maintenance procedures, operators should migrate the VIPs off the affected node first.

BZ#1650702

This update fixes a configuration issue that caused failure of operations on volumes that use Nova's privileged API (for instance, migrating an in-use volume). 

The failures happened because the OpenStack Platform director was not configuring authentication data required for the Block Storage service (Cinder) to access privileged portions of the Nova API.

The director now configures Cinder with Nova's authentication data. As a result, operations on volumes that require privileges succeed.

BZ#1623554

This update adds a TripleO heat template parameter as an option for setting RX/TX queue size. 

Prior to this update, users could set RX/TX queue size with `nova::compute::libvirt::rx_queue_size/nova::compute::libvirt::tx_queue_size`. However, there was no dedicated TripleO heat template parameter.
With this update, users can set the RX/TX queue size either on a global level using:

```
parameter_defaults:
NovaLibvirtRxQueueSize: 1024
NovaLibvirtTxQueueSize: 1024
```

or overwrite the hieradata via [ROLE]ExtraConfig, which then can be used to configure a subset of compute nodes for which a dedicated role was created:

```
parameter_defaults:
NovaComputeExtraConfig:
	nova::compute::libvirt::rx_queue_size: 1024
	nova::compute::libvirt::tx_queue_size: 1024
```

Note: The possibilities described above are mutually exclusive.

os-net-config

BZ#1654987

This update ensures that VLAN interfaces are restarted when underlying devices are restarted after a device configuration change, allowing the successful restoration of networks.

Prior to this update, a VLAN interface was not restarted when the underlying device was restarted. Network routes using the VLAN interface as the next hop were removed and not restored.

With this update, the VLAN interfaces are restarted when the underlying devices are restarted. Network routes are restored.

puppet-tripleo

BZ#1649363

This update lets the operator specify custom timeouts for each haproxy back end via special hiera keys:

ExtraConfig:
tripleo::haproxy::cinder::options:
	'timeout client': '90m'
	'timeout server': '90m'

With this support an operator can specify custom options for each haproxy back end.

python-os-brick

BZ#1599641

This update fixes an issue that sometimes prevented detachment of multipath devices. 
Prior to this update, detachment of multipath devices included a flush of each individual path. If an individual path flushing failed, the detachment failed, even though there were other paths available to flush all the data. 

Because flushing the multipath already ensures buffered data is written on the remote device, individual paths are no longer flushed. As a result, detaching only fails when the detachment would actually result in data loss.

BZ#1583466

iSCSI device detection checked for the presence of devices based on the re-scan time. Devices becoming available between scans went undetected. With this release, searching and rescanning are independent operations working at different cadences with checks happening every second.

BZ#1634163

Under certain circumstances, the os-brick code responsible for scanning FibreChannel HBA hosts could return an invalid value. The invalid value would cause services such as cinder and nova to fail. With this release, the FibreChannel HBA scan code always returns a valid value. Cinder and nova no longer crash when scanning FibreChannel HBA hosts.